Retrosplenial and postsubicular head direction cells compared during visual landmark discrimination:

Yave Roberto Lozano; Hector Page; Pierre-Yves Jacob; Eleonora Lomi; James Street; Kate Jeffery

doi:10.1177/2398212817721859

Retrosplenial and postsubicular head direction cells compared during visual landmark discrimination:

Lozano, Yave Roberto; Page, Hector; Jacob, Pierre-Yves; Lomi, Eleonora; Street, James; Jeffery, Kate 2017-09-15 00:00:00 Background: Visual landmarks are used by head direction (HD) cells to establish and help update the animal’s representation of head direction, for use in orientation and navigation. Two cortical regions that are connected to primary visual areas, postsubiculum (PoS) and retrosplenial cortex (RSC), possess HD cells: we investigated whether they differ in how they process visual landmarks. Methods: We compared PoS and RSC HD cell activity from tetrode-implanted rats exploring an arena in which correct HD orientation required discrimination of two opposing landmarks having high, moderate or low discriminability. Results: RSC HD cells had higher firing rates than PoS HD cells and slightly lower modulation by angular head velocity, and anticipated actual head direction by ~48 ms, indicating that RSC spiking leads PoS spiking. Otherwise, we saw no differences in landmark processing, in that HD cells in both regions showed equal responsiveness to and discrimination of the cues, with cells in both regions having unipolar directional tuning curves and showing better discrimination of the highly discriminable cues. There was a small spatial component to the signal in some cells, consistent with their role in interacting with the place cell navigation system, and there was also slight modulation by running speed. Neither region showed theta modulation of HD cell spiking. Conclusions: That the cells can immediately respond to subtle differences in spatial landmarks is consistent with rapid processing of visual snapshots or scenes; similarities in PoS and RSC responding may be due either to similar computations being performed on the visual inputs, or to rapid sharing of information between these regions. More generally, this two-cue HD cell paradigm may be a useful method for testing rapid spontaneous visual discrimination capabilities in other experimental settings. Keywords Head direction cells, postsubiculum, retrosplenial cortex, landmarks, visual discrimination, spatial memory, in vivo rodent electrophysiology Received: 17 April 2017; accepted: 28 June 2017 Introduction How the brain forms a representation of external, navigable space same relative PFDs, even if the orientation of the entire cell popula- is a current area of intense enquiry because it involves transforma- tion changes from one environment to the next. Local environmen- tion of sensory inputs into higher-order, more abstract cognitive tal landmarks – mainly visual – establish the population orientation structures, and thus has wide relevance to cognition generally. when the animal enters an environment, and the signal is updated as One of the foundations of the place representation is the ‘sense of the animal moves around by means of self-motion information such direction’, supported by a network of brain regions known as the as vestibular, optic flow, motor efference and proprioceptive cues to head direction (HD) system, which uses previously learned visual movement (Taube, 2007; Yoder et al., 2011). Although HD cells landmarks to re-orient when an animal re-enters a familiar envi- typically begin firing essentially immediately on entry into a famil- ronment. This study investigates the neural basis of this rapid ori- iar environment (Jankowski et al., 2014), the environmental cues entation process, which is important for understanding how perception and memory processes shape the place representation. Division of Psychology and Language Sciences, University College The HD system in rodents (and probably all vertebrates) con- London, London, UK tains so-called HD cells, which fire when the animal faces in a par- ticular direction, regardless of position, and are close to silent Corresponding author: otherwise (Taube et al., 1990a; 1990b). Each cell has its own pre- Kate Jeffery, Division of Psychology and Language Sciences, University ferred firing direction (PFD) in a given environment, and the College London, 26 Bedford Way, London WC1H 0AP, UK. ensemble of HD cells is coherent – that is, the cells maintain the Email: k.jeffery@ucl.ac.uk Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License (http://www.creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage). 2 Brain and Neuroscience Advances are learned, not hard-wired (Taube and Burton, 1995), which means PoS rat had received a saline sham injection into the lateral genicu- that some kind of rapid recognition process must take place. How late nucleus (LGN), as part of a different study. The rats were housed and where this occurs is not known, but is likely to be in HD regions in individual cages in a temperature and humidity-controlled colony close to the visual system. room that was kept on an 11:11 h light:dark cycle plus 1 h each of Current evidence suggests that the postsubiculum (PoS) and the half-light simulated dawn and dusk. All the rats had free access to retrosplenial cortex (RSC) have an important role in processing food and water prior to surgery and began food restriction to main- landmark information derived from visual areas and relaying this tain 90% of their free-feeding weight 1 week after surgery. All proce- input to interconnected areas of the HD cell circuit and the hip- dures were licensed by the UK Home Office following the revised pocampal formation (for review, see Yoder et al., 2011). Consistent ASPA regulations (2013) modified by the European Directive with this hypothesis, Goodridge and Taube (1997) found that tuning 2010/63/EU. Two of the RSC-implanted rats also took part in curve precision (width of the tuning curve) and stability of anterior another experiment, in a different apparatus (Jacob et al., 2016). thalamic HD signals were impaired after PoS lesions, and the cells showed reduced responsiveness to rotations of the landmarks. Yoder et al. (2015) reported a similar effect on lateral mammillary HD neu- Electrodes and surgery rons, as did Calton et al. (2003) on CA1 place fields, downstream of Each rat was anesthetised and implanted with a microdrive the HD signal. Similarly, Clark et al. (2010) found the same effects (Axona) that was configured with either four or eight tetrodes on both tuning curve width/stability and landmark control following threaded inside a guide cannula. Each tetrode was made of indi- RSC lesions, although to a lesser extent. In contrast, lesions of the vidual 90:10 platinum–iridium wires (California Fine Wire) of parietal (Calton et al., 2008) or the postrhinal (Peck and Taube, diameter 25 µm (for the four-tetrode drives) or 17 µm diameter 2017) cortex did not impair landmark control of anterior dorsal thal- (for the eight-tetrode drives). The tips of the electrodes were amus (ADN) HD cells, suggesting that the processing of visual land- plated in a 1:9 0.5% gelatine:Kohlrausch platinum solution to mark information is routed via the PoS and the RSC. These studies approximately 250 kΩ using a pulse generator (Thurlby Thandar collectively point to a likely role for PoS and RSC in passing infor- TGP-110) and a current source amplifier (A.M.P.I. ISO-FLEX) mation about landmarks (or at least their visual properties) to the that delivered 2 µA current for 550 ms to each channel. The subcortical circuits that collate the HD information and generate a microdrive and the guide cannula were fastened to the skull with stable signal. It is thus of interest to look for possible differences in dental acrylic (Simplex Rapid) covering seven supporting the contributions made by the two regions. screws of 1.6 × 3 mm (Small Parts) that were inserted into the It is not known why there should be two separate landmark pro- occipital, parietal and frontal cranial bones. One of the support- cessing regions, but one possibility is that they differ in some aspect ing screws made contact with the frontal cortex and was con- of their contribution to the processing of environmental landmarks. nected to the microdrive ground wire. We therefore set out to compare the responses of these neurons dur- Implant coordinates were based on previous studies of HD ing re-orientation in a situation where detailed landmark-cue pro- cells in the RSC (Cho and Sharp, 2001) and the PoS (Taube cessing is required. We recorded HD neurons as rats foraged in a et al., 1990a); for RSC (n = 12 rats; n = 6 left hemisphere, n = 6 cylinder within a curtained enclosure (Figure 1(a)). The infinite rota- right hemisphere), these were 5.4 mm posterior to bregma, 0.6– tional symmetry of this arrangement (Figure 1(b)) was broken by a 0.8 mm lateral to the midline and 0.2–0.9 mm ventral from the pair of cue cards on the cylinder wall, located opposite each other; cortical surface, while two sets of coordinates were used for the the resulting twofold symmetry of the cue pair (Figure 1(c)) was PoS (n = 5 rats; all in the left hemisphere); being, respectively, further broken by making the cards different so that the environment 6.7 or 7.5 mm posterior to bregma, 2.8 or 3.2 mm lateral to the was now polarised, provided the cues could be discriminated (Figure midline and 1.6 or 1.9 mm ventral from the cortical surface. 1(d)). The discriminability of the cue card pair varied between high, After surgery, the animals were monitored until they awoke, and moderate and low (Figure 1(e)) – in the HIGH condition, one card meloxicam (Metacam) was given in jelly for three consecutive was black and one white; in the MOD condition, a black bar was days as pain relief. All animals were allowed to recover for oriented and/or positioned one way on one card and the other way on 1 week prior to the start of recording. the other, and in the LOW condition, the cards were visually identi- cal (although they may have been distinguishable by smell; hence discriminability is assumed to be low and not zero). Cue control apparatus PoS and RSC HD cells were recorded in sessions ranging from 4 to 12 trials; between trials, the rat was removed and mildly The cue control experiments were performed inside a cylindri- disoriented, and the entire environment was frequently also cal arena (diameter, 74 cm; height, 50 cm) made of plywood rotated to disconnect it from static external cues. Cells were painted with light grey matt acrylic and placed at the centre of tested for basic firing parameters (firing rate, spiking characteris- a black curtained enclosure (diameter, 260 cm) (Figure 1(a)). A tics, and possible spatial localisation) and to determine whether cylinder was selected as a recording arena to minimise the their firing directions rotated appropriately with the cue pair, influence of the environment’s geometry as an orienting cue indicating successful detection and discrimination. (Golob et al., 2001; Knight et al., 2011). Attached to the inner wall of the cylinder with Velcro tape were two 50 × 50 cm cue cards (Figure 1(b)) made from black and/or white polypropyl- Materials and methods ene sheets, each subtending ~77° of arc, and located 180° apart. Two of the cue pairs were plain – either both black (iden- Subjects tical-cue controls; B-B) or one black and one white (a maxi- A total of 18 adult male Lister Hooded rats weighing between 317 mally salient high-contrast condition; B-W). The remaining and 437 g at the time of surgery were used for the experiments. One cue pairs, with one exception, were made from white card Lozano et al. 3 Figure 1. Cue-discrimination setup for recording HD cells in the RSC and PoS of freely moving rats. (a) Recording environment. Rats foraged for rice in a 50 cm high by 74 cm diameter cylindrical arena with two cue cards 50 × 50 cm attached to the inner wall 180° apart. The arena was situated in the centre of a 260 cm diameter, black circular curtained enclosure. (b)–(d) Environmental symmetry, and hence HD cell anchoring, varies as a function of cue perception. (b) If no cues are detected then the environment has infinite rotational symmetry, and HD orientations might be random. (c) If the cue pair is detected by the cells but not discriminated, then the environment has twofold rotational symmetry and an HD cell might fire in either of two directions, at random. (d) If the cues are both detected and discriminated, then the environment lacks rotational symmetry (is polarised) and an HD cell would be able to fire in a constant direction. (e) Cue card patterns, grouped into discriminability categories (high, moderate and low), and their corresponding abbreviations. B-W: black–white, V-H: vertical–horizontal, L-R: left–right, T-B: top–bottom, B-B: black–black and L-L are both L-shaped. The V-H cards were also sometimes used in white-on-black configuration. decorated with a black bar of 14 × 50 cm; these were thus equal the exceptional pair, black and white were reversed for the ori- in overall luminance and contrast, so discrimination would entation cues. require processing of the cues’ internal structure. The position The arena rested over a black vinyl sheet and was lit from of the bar in each pair differed in, respectively, orientation above by six light fixtures that provided approximately 250 lux (vertical vs horizontal; V-H cues), lateral position (left vs right; of light, with a radio attached to the ceiling as a source of white L-R cues) or vertical position (top vs bottom; T-B cues). For noise to reduce the effect of directional auditory cues. 4 Brain and Neuroscience Advances encourage the rat to sample all the locations and facing directions Screening/recording procedures within the cylinder. With the exception of the black–white (B-W) Signal processing and tracking. Single units were recorded cue pair, which was used to establish cue control, the order of using a headstage amplifier connected to a microdrive and con- presentation for the different cue card stimuli was pseudo-ran- nected to the multichannel recording system (DacqUSB, Axona) domised between rats to prevent temporal effects due to the via a flexible lightweight tether. Traces from individual channels changing experience of the animal across days. were collected at a sampling rate of 48 kHz, amplified 6000– For one rat, a control procedure was conducted to assess the 20,000 times and band-pass filtered from 300 Hz to 7 kHz. The relative contribution of vision and olfaction to the landmark dis- traces from each electrode were referenced against an electrode crimination. Using the patterned cues that had rotational symme- from a different tetrode that showed low spiking activity. Spikes try (V-H, T-B and L-R), the visual identity of the cards was were defined as short-lasting events that crossed a user-defined reversed by rotating around their central points, 90° for V-H and threshold; the period from 200 µs before to 800 µs after threshold- 180° for T-B and L-R, to determine whether firing remained crossing were captured and saved along with the corresponding aligned relative to the physical cards or to the visual appearance. timestamps. To track the rat’s head position and facing direction in the horizontal plane, two arrays of light-emitting diodes (LEDs), Data analysis one large and one small, were transversely positioned 8 cm apart on a stalk connected to the headstage. The LED positions were Spike-sorting was performed using KlustaKwik (Kadir et al., recorded at a sampling rate of 50 Hz by a camera attached to the 2014) followed by manual refinement using the TINT software ceiling. Spike times, the rat’s head position in x- and y-coordi- package (Axona). Polar plots were generated, and cells that nates and its heading in degrees were saved for offline analysis. showed directional firing by eye were selected for further quanti- tative analysis, which included extracting peak firing rate and directional clustering – Rayleigh vector (R-vector) – scores. HD cell screening. The screening sessions for single units tuned Cells that were recorded on the same tetrode across sessions were to HD were conducted inside a 76 × 76 × 50 cm box in a room treated as unique if more than 5 days separated the sessions. For separate from where the cue control experiments took place. The each trial, a cell was accepted into further analysis if it had a peak animals had visual access to a polarising cue card placed inside firing rate >1.0 Hz across sessions, a peak:mean rate ratio > 2 and the box, and distal room cues. The screening sessions consisted a R-vector p-value <0.05. Cells were excluded altogether if only of 5–10 min trials during which the rat foraged for rice while three trials or fewer per session remained after selection. spikes were monitored. Polar plots were generated with spike- Data were grouped in three ways: (a) Those pertaining to sorted data using the software TINT (Axona) and examined to cell-specific characteristics, such as firing rate, R-vector score determine whether single units were tuned to HD. In screening and tuning curve width were averaged across a given session. sessions during which HD cells were not found, the tetrodes were (b) Data concerning trial-specific characteristics such as firing lowered by ~50 µm and the rat was screened again 4 h later or the direction relative to the cue card array were averaged across a given next day. cell. (c) Data concerning session-specific characteristics, such as the spread of firing directions, were averaged across a session. Cue control procedure. After isolating single units tuned to HD The basic protocol for firing direction analysis is shown in Figure during screening, the rat was taken to the recording room inside a 2. First, a tuning curve was derived (see below). Each tuning curve closed opaque box. A cue control session was conducted using a was recorded in the camera frame of reference, within which the similar recording procedure as described previously (Knight cues rotated, but it was necessary to align these within a common et al., 2011). Briefly, each session started with a series of baseline reference frame so that population statistics could be derived. To do trials (2–4 trials) where the two cue cards remained aligned in the this, the tuning curves were first specified relative to one of the two same location relative to the room coordinates. This was followed cues (‘Cue 1’; Figure 2(b)) and the mean firing direction across the by a series of rotation trials (4–8 trials) during which the cue cards session computed. Then, the tuning curves were realigned relative to were rotated together by ±45°, ±90°, ±135° or 180° (Figure 1(c)) this mean; this value (deviation from the session mean) became the to test whether HD cells could discriminate and use the visual value that entered into the population analysis. Finally, the angles features of the cue cards as orienting landmarks (Figure 1(d)). For were doubled in order to remove any bipolarity that might be present each recording session, the starting location of the cue cards as due to cue confusion – this was done so as to enable determination well as the magnitude and direction of the cue rotations were of cue use independently of cue discrimination. pseudo-randomised. The length of each trial was 300 s and was initiated via remote control after placing the rat inside the cylinder with a pseudo-random location and facing direction. Prior to each Cell-specific firing characteristics trial, the cue cards, recording arena and the base of the cylinder were inverted and wiped with 75% ethanol to scramble olfactory Data were analysed using MATLAB (MathWorks) with cus- cues. During the inter-trial interval, the rats remained inside a tom-made programs and functions taken from the CircStat tool- holding box outside the curtained enclosure, and then prior to box (Berens, 2009). To analyse the firing characteristics of being replaced in the recording box they were mildly disoriented single units, the spike times and position samples of the rat’s by being passively transported and rotated in the holding box facing direction were sorted into bins of 6°. The mean firing around the periphery of the curtained enclosure, thereby prevent- rate per HD bin (Hz) was calculated by taking the sum of the ing the animals from using self-motion cues to track their orienta- spikes divided by the total amount of time (s) that the rat’s fac- tion. During the recording trials, the experimenter remained ing direction was located in each bin. A 5-bin (30°) smoothing outside the curtained enclosure, tossing rice into the arena to kernel was applied to the circular histogram of the firing rate as Lozano et al. 5 Figure 2. Tuning curve analysis. (a) Example of the tuning curve derived from a single PoS HD neuron in a series of recording trials, selected from a 12-trial session. The plots show firing rate as a function of HD, normalised to the peak rate (shown). (b) Schematic from a hypothetical set of trials showing how data were transformed into a common reference frame to allow population comparisons. The original reference frame was that of the room-fixed camera: the top row shows an idealised trial in which the cell consistently rotated its firing with the cue pair, but was aligned relative to Cue 1 on trials 1, 2 and 5 and 180° away from it on trials 3 and 4. The bottom row shows these data expressed in the different reference frames: in the camera frame, the firing direction is scattered; relative to Cue 1, it is bipolar with three trials in one direction and two in the opposing direction; relative to the session mean (arrow), the bulk of the firing is now aligned to zero, and after angle-doubling the bipolar distribution is now transformed to unipolar. a function of HD to minimise random influences to the firing the effect of wrapping the 180° points around to zero, thus render- rate in each cell (Abeles, 1982). Polar plots of the smoothed ing a bipolar distribution unipolar for the purposes of analysis. histograms were generated to visualise the cell’s tuning curve (Figure 2(a)). Test for bipolar tuning curves. In order to test whether there The parameters that were used to quantify the tuning curve was any within-trial tendency towards bipolarity, due to (per- characteristics for each cell were firing rate, PFD, tuning width, haps) intermittent reversing of the tuning curve arising from con- and mean R-vector length. The firing rate of the HD cell was fusion between the cues, a test for bipolarity was conducted by defined as the bin with the maximum firing rate; that is, the mode running a circular autocorrelation on the smoothed binned data of the circular histogram, and the PFD defined to be the direc- from each trial, using the MATLAB circshift and corr functions. tional bin of the mode. The tuning width, or directional firing From the autocorrelation, the values at 90° and 270° relative to range of the cell, was determined by computing two standard the peak were extracted and averaged, and compared with the deviations from the circular mean direction. The R-vector score peak at 180° using a one-tailed t-test of the hypothesis that the was used as a measure of the directional tuning in each cell. 180° peak would be larger. Values of R close to 1 indicate that the spikes are closely clus- tered around a single value and values close to 0 indicate that the spikes were distributed around all facing directions. Spatial firing patterns. In order to determine whether there To examine how a cell behaved in relation to the cue card pair was any spatial specificity to the firing, which could be of com- across trials, the PFDs were realigned according to their relation- putational utility (Bicanski and Burgess, 2016), and has been ship to one of the two cues (arbitrarily called ‘Cue 1’) and the reported for some types of HD cells in cortical regions (Cacucci circular mean of the session values was computed. Each PFD et al., 2004; Peyrache et al., 2016), we examined the spatial dis- was then expressed as a deviation from this mean. tribution of spikes in the same way as is usually done for place In some cases, it was necessary to test whether a low R-vector cells, extracting the spatial information content of the firing, the score could be due to bipolarity of the firing distribution. To do coverage relative to the whole environment and the coherence of this, the values were doubled and plotted modulo 360 – this has the firing distribution. Path data were extracted by smoothing the 6 Brain and Neuroscience Advances position points with a 400 ms window and first plotting the spikes compensate for variability in firing rate between cells. Running at their corresponding locations, for visual inspection. These data speeds below 2 cm/s were excluded from running speed analy- were then used to generate dwell-time-normalised firing rate sis. Linear running speed was binned in intervals of 2 cm/s, and maps by binning the spike and position data into 3 cm bins and AHV was binned in intervals of 2°/s. The firing rate was calcu- dividing the firing rate by dwell time, and then smoothing the lated by counting the spikes in each bin and dividing by the time map with a boxcar of three spatial bins (whereby the value in spent in that bin (dwell time) and then normalised to the peak for each bin was replaced by the average for that bin plus the sur- that trial to enable comparison across trials/cells. Bins with rounding eight bins). To eliminate sampling bias due to the rat’s dwell times of less than 1.5% total trial time or with fewer than inability to face all directions everywhere in the arena, we under- five spikes were discarded; a linear regression was run on the took the spatial analysis using only the inner 50% of the arena, in remainder to generate a slope value. Because dwell time which directional sampling was homogeneous. In case some cells decreased with increasing velocity, which might cause artefacts might have place fields near the edge of the arena, we also anal- in the rate/speed relationship, the baseline for each trial was cal- ysed the hemi-cylinder lying in the direction of the cell’s PFD, culated by generating an artificial continuous 10 Hz spike train, for which the directional bias was much reduced (since it was analysing it in the same way and subtracting this control slope easy for the rat to face in the cell’s PFD in this region); this made from the raw data slope. For AHV, left and right turns were ana- little difference, so we chose the inner region as being the most lysed separately and the absolute slope values then combined for conservative measure. that cell. This is because previous recordings from other brain Spatial information in bits/spike was computed following the regions have found cells with asymmetric AHV rate profiles method of Skaggs et al. (1993). Coverage was calculated as the (Bassett and Taube, 2001), being negative in one direction and percentage bins above 20% peak rate, and coherence of spatial positive in the other, which would cancel if the raw values were firing was determined using a Pearson’s correlation between the taken. The resulting difference values were entered into a t-test smoothed and unsmoothed rate maps. Actual data were com- comparing PoS and RSC. pared against a control dataset generated by taking the spike data Anticipatory time intervals (ATIs) were estimated using a time and shifting it forwards by 1000 position points (20 s), which slide analysis in the manner of Blair and Sharp (1995). Given the would scramble any spatial specificity of firing while leaving the camera sample rate of 50 Hz, spike times were shifted forwards in temporal dynamics unchanged. Each firing rate map was com- 20 ms intervals from 20 to 160 ms. For each head turn, a tuning pared with its shuffled control map using a paired t-test. curve was constructed for leftwards and rightwards head turns in the manner described. A population vector method was used to calculate the PFD of each tuning curve (Song and Wang, 2005) as Temporal firing patterns. We looked at temporal patterns of firing including inter-spike interval (ISI) time to peak and decay time to half-peak, as well as theta-frequency rhythmicity. For the   ISI analysis, only trials with >145 spikes were used and only one rx sin( ) ii   PFD = arctan trial (usually the first) was used from each cell. A histogram of  N   rx cos( )  ISIs with 2 ms bins was generated for each cell and the peak was ii  i  taken as the centre of the bin with the highest count. We calcu- lated decay time by fitting, to the histogram, a one-term exponen- bx where r is the firing rate in bin i of N with mean bin direction x , and i i tial decay function of the form ya = from the peak to peak + 1 s, arctan denotes quadrant-specific arctangent function. using the fit function from MATLAB’s Curve Fitting toolbox. The difference between PFD for leftwards and rightwards Time to half-peak was then taken as the time taken for the expo- tuning curves was plotted, and a line was fitted to it using the nential fit to decay to half the peak value. MATLAB function polyfit. To allow for estimations of ATIs at a We measured theta modulation by plotting autocorrelograms finer scale than measurement, the ATI was taken as the value of of the spike trains over the range ±500 ms, in bins of 10 ms dura- this fitted line when PFD difference was equal to zero, extracted tion. The plots were then highly smoothed (20 bins) to remove using the MATLAB function polyval. local variations, and the values at the 7th bin from the central peak (expected trough at 60–70 ms) and the 12th bin (expected peak at 120–130 ms) determined: the theta modulation index was Stabilisation analysis taken as the difference between these values divided by their To investigate the time course of cue control establishment, we sum. If there is significant theta modulation, then the 12th bin looked at firing within the PFD range (PFD ±20°) across the trial, should be a peak and the 7th bin a trough, yielding a positive taking both the percentage of total spikes emitted in each time modulation index varying from 0 to 1. Conversely, values below decile (30 s – trials were 300 s long), and the time taken to reach zero would indicate a descending likelihood of a cell spiking with each spike count decile. time between the first and second time-points. Movement correlates. The relationship between linear or Histology angular speed and firing rate was examined by analysing those portions of the trial when the animal’s HD was within 45° either After completion of the electrophysiological recordings, the rats side of the PFD of the cell, and correlating the firing rate with were anaesthetised and killed with an overdose of sodium pento- movement speed. Correlations of firing rate with linear running barbital Euthatal, 150 mg/kg and perfused transcardially with speed and angular head velocity (AHV) were computed as per- saline followed by 4% formalin solution. A day before section- centage firing rate change as a function of movement speed, to ing, the brains were placed in 4% formalin/20% sucrose solution Lozano et al. 7 Table 1. Basic firing statistics compared between PoS and RSC HD cells. PoS (n = 74) RSC (n = 75) F/T statistic (dof) p value −7 Peak rate (Hz) 8.14 ± 0.63 22.40 ± 2.58 t(147) = 5.34 3.48 × 10 −8 Mean rate (Hz) 1.83 ± 0.19 5.07 ± 0.52 t(147) = 5.86 2.92 × 10 −5 Tuning curve width (°) 43.65 ± 1.13 49.29 ± 1.28 t(147) = 3.31 1.18 × 10 Tuning curve R-vector 0.59 ± 0.02 0.53 ± 0.02 t(147) = 1.53 0.13 −7 Spatial information (data: control 1.63 ± 0.06 2.22 ± 0.10 t(145) = 5.00 7.00 × 10 bits/spike ratio) −4 Coverage (data: control % ratio) 0.98 ± 0.01 1.48 ± 0.04 t(145) = 3.67 1.70 × 10 −4 ISI rise time to peak (ms) 13.5 ± 1.6 7.5 ± 0.4 t(144) = 3.74 2.66 × 10 −4 ISI decay time to half-peak (ms) 80.1 ± 6.5 48.8 ± 5.8 t(144) = 3.61 4.20 × 10 Theta modulation index –0.05 ± 0.03 –0.03 ± 0.00 t(148) = 2.64 0.009 Abs. slope of firing rate correlation 0.69 ± 0.1 vs 0.06 ± 0.001 0.6 ± 0.1 vs 0.07 ± 0.01 Data vs 10 Hz control < 0.0001 with linear running speed (%/m/s) F(1, 1) = 40.10 vs 10 Hz control PoS vs RSC 0.50 F(1, 1) = 0.50 Interaction 0.56 F(1, 64) = 0.30 Abs. slope of firing rate correlation 0.12 ± 0.01 vs 0.03 ± 0.05 0.08 ± 0.01 vs 0.03 ± 0.03 Data vs 10 Hz control 0.003 with angular head velocity (%/°/s) F(1, 1) = 9.70 vs 10 Hz control PoS vs RSC 0.003 F(1, 1) = 5.0 Interaction 0.025 F(1, 64) = 4.20 Anticipatory time interval (ms) 14.37 ± 8.19 47.91 ± 4.28 t(147) = 3.62 0.0002 PoS: postsubiculum; RSC: retrosplenial cortex; dof: degrees of freedom: ISI: inter-spike interval. N’s differ slightly because some of the data trials could not be curve-fitted. for cryoprotection. Coronal or sagittal sections of the brains were granular RSC as illustrated by the final tetrode depth, which cut in 40 µm sections using a microtome at −20° C and mounted could have under-estimated the actual number of HD cells. on microscopic slides (Superfrost BDH). The brain sections were Results and statistics are detailed below and summarised in Nissl stained with 0.1% cresyl violet (Sigma-Aldrich) or 0.5% Table 1. thionin (Sigma-Aldrich) and cover-slipped with DPX (Sigma- Aldrich). The slides were examined under a light microscope Cell-specific firing parameters (Leica) and imaged with a digital camera mounted on the micro- scope. Images with visible electrode tracks were saved for histo- As described in section ‘Methods’, the cell-specific character- logical analysis. The electrode location was verified by examining istics of peak and mean firing rate, R-vector score and tuning the brain region where the final electrode track was located and curve width were calculated for each trial and then averaged then relating the location to a standard rat atlas of the regions across a given session for each cell. Statistical parameters are (Paxinos and Watson, 2007). summarised in Table 1 and graphically shown in Supplementary Figure 2(A). Peak firing rates were significantly higher in RSC, being 8.14 ± 0.63 Hz in PoS and 22.40 ± 2.58 Hz in RSC Results −7 (t(147) = 5.34, p = 3.48 × 10 ). Similarly, mean firing rates were higher in RSC: 1.83 ± 0.19 in PoS and 5.07 ± 0.52 in RSC Representative example sections showing the final tetrode −8 (t(147) = 5.86, p = 2.92 × 10 ). Tuning curves were narrower placement of rats implanted in the PoS and the RSC are shown in PoS, being 43.65° ± 1.13° in PoS and 49.29° ± 1.28° in RSC in Supplementary Figure 1. A total of 149 unique cells meeting −5 (t(147) 3.31, p = 1.18 × 10 ). However, directional tuning was the criteria for HD cells (see section ‘Methods’) were recorded similar in both structures: there was no difference in R-vector from 18 rats in 78 sessions, of which 74 cells were recorded score, being 0.59 ± 0.02 in PoS and 0.53 ± 0.02 in RSC from 6 rats implanted in the PoS over 34 sessions, and 75 cells (t(147) = 1.53, p = 0.13). from 12 rats implanted in the RSC over 44 sessions. We observed that a greater proportion of RSC HD cells (73%) were found below 1500 µm which is likely in the deeper granu- Spatial firing lar region. However, although the implant target coordinates were the same for all rats, there was a degree of variability in Visual inspection of the firing rate maps revealed mostly uni- the actual anterior-posterior and medio-lateral coordinates in form firing, although there were sometimes patches of inhomo- which the tetrodes were implanted (Supplementary Figure 1). geneous firing at a single-trial level and occasional clear Furthermore, the tetrodes did not cover all the layers of the place fields of the kind seen in hippocampal recordings 8 Brain and Neuroscience Advances (Supplementary Figure 3). We quantified spatial modulation by Consistent with our observations of higher firing rates in RSC, the considering spatial information content, coverage and coher- time to the ISI peak was considerably longer in PoS (13.5 ± 1.6 ms) ence. As detailed in section ‘Methods’, only the central 50% (by than in RSC (7.5 ± 0.4 ms), these being significantly different −4 radius) of the arena was used in order to remove bias induced by (t(144) = 3.74, p = 2.66 × 10 ). The time to return to the half-peak, inhomogeneous directional sampling around the arena edges. which can be thought of as a measure of the prevalence of longer Both PoS and RSC HD neurons showed evidence of a spa- ISIs, was also different, being 80.1 ± 6.5 ms for PoS and 48.8 ± 5.8 ms −4 tial component to their firing, revealed by higher spatial infor- for RSC (t(144) = 3.61, p = 4.2 × 10 ). mation content, lower coverage and by higher coherence For each cell, autocorrelograms were generated for each trial relative to the time-shifted control data (Supplementary Figure in a session (Supplementary Figure 5(A)) and from these, a theta 2(B)). Spatial information content (bits per spike) for the PoS modulation index was calculated (see section ‘Methods’). Visual data was 0.23 ± 0.02, while for the control data it was inspection of the autocorrelograms revealed no evidence of theta 0.16 ± 0.01, which was significantly different, as revealed by a modulation, although there were frequently peaks and troughs −18 paired one-tailed t-test (t(73) = 11.57, p = 9.72 × 10 ). suggestive of slightly longer scale periodicity (see examples Corresponding values for the RSC data were 0.15 ± 0.01 and from the cell shown in the Supplementary Figure 5(A)). However, for the control data were 0.07 ± 0.00, which was significantly these patterns did not persist across trials and are likely due to the −17 different (t(73) = 11.16, p = 1.13 × 10 ). Coverage was slightly dynamics of the animal’s movements. For example, if a rat is less for both cell types in the real data; PoS coverage in the real foraging by sweeping its head back and forth, then a cell’s tuning data condition was 52 ± 3%, and in the time-shifted control curve will be visited and re-visited over a regular time period of condition was 53 ± 3% which was significantly greater (paired up to several seconds. It may be that rats varied their movement −5 one-tailed t(72) = 4.14, p = 4.70 × 10 ). For RSC, coverage for patterns across trials. the data was 62 ± 2% and for the control data was 64 ± 2%, The quantitative theta modulation index confirmed the visual −4 which was also significantly higher (t(73) = 3.58, p = 3.1 × 10 ). impression: in general, values were, if anything, negative (a Coherence values were greater for the data compared with the lower value at the expected peak than at the expected trough) and time-shifted control data. For the PoS data were 0.24 ± 0.01 slightly more so, and more dispersed, for PoS (Supplementary and for the time-shifted control data were 0.20 ± 0.01, which Figure 5(B)). The values for PoS were −0.049 ± 0.006 and for −9 was significantly lower (t(70) = 6.38, p = 8.4 × 10 ), while for RSC were −0.032 ± 0.002 (t(148) = 2.63, p = 0.005). RSC the real data had a coherence value of 0.27 ± 0.01 and the control data had a coherence of 0.19 ± 0.00, which was again −19 Movement correlates significantly lower (t(73) = 12.01, p = 2.89 × 10 ). We then compared spatial firing bias between PoS and RSC Correlations of firing rate with linear running speed and AHV by computing the ratio of data to control values for each ses- were computed as described in section ‘Methods’, yielding per- sion and then comparing between cell types for each of the centage firing rate change as a function of movement speed. For three spatial parameters. For spatial information, the ratio of linear running speed, there was a weak correlation of firing rate data: control was 1.63 ± 0.06 for PoS and 2.22 ± 0.10 for RSC, with running speed relative to the control 10 Hz spike train; this −7 which was significantly different (t(145) = 5.0, p = 7 × 10 , was 0.69 ± 0.1%/m/s for PoS and 0.61 ± 0.1%/m/s for RSC. A two- d = 0.83). For coverage, the ratio for PoS was 0.98 ± .01 and for way analysis of variance (ANOVA) of data type (real vs 10 Hz RSC was 0.99 ± .01, which did not differ (t(145) = 0.67, control) and brain area found a main effect of data (F(1, 1) = 40.1, p = 0.25). For coherence, the ratio for PoS was 1.26 ± 0.04, and p < 0.0001), but no effect of brain area (F(1, 1) = 0.50, p = 0.5) and for RSC it was 1.48 ± 0.04, which was significantly different no interaction (F(1, 64) = 0.30, p = 0.56). −4 (t(145) = 3.67, p = 1.7 × 10 , d = 0.61). For AHV, there was also overall a very weak relationship Overall, then, both PoS and RSC HD neurons showed a with firing rate relative to the control steady 10 Hz spike train; small amount of spatiality to their firing when compared with this was 0.12 ± 0.01%/°/s for PoS and 0.08 ± 0.01%/°/s for a time-shifted version of the same data, having higher spatial RSC. These effects, though small, were significant: a two-way information and spatial coherence; this was more pronounced ANOVA comparing data versus control slopes for PoS and in RSC. This accords with the visual inspection showing occa- RSC found a main effect of data type (F(1, 1) = 9.70, p = 0.003), sional and reproducible spatial inhomogeneity of firing. It thus a main effect of brain area (F(1, 1) 5.00, p = 0.03) and a sig- appears that there is a degree of spatial modulation of firing, nificant interaction (F(1, 64) = 4.20, p = 0.025). Plots of firing but this was rather slight. For RSC, this is surprising given rate against HD and AHV (Figure 3) revealed an additional previous reports of conjunctive spatial and directional firing in difference between PoS and RSC neurons. An example of this structure (Cho and Sharp, 2001). However, it is consistent each (the closest cell to the mean in each case) is shown in with our previous studies of RSC HD cells, in which we have Figure 3(a), where it can be seen that the cell’s directional tun- observed very little clear spatial firing (Jacob et al., 2016; ing deviates as a function of AHV, with greater deviation for Knight et al., 2014). higher AHVs in the RSC but not the PoS neuron. This devia- tion reflects anticipatory firing, first reported by Sharp and colleagues for anterior thalamic HD neurons but not PoS Temporal components of firing (Blair et al., 1997; Blair and Sharp, 1995) and replicated by Taube and Muller (1998); anticipatory firing was subse- To analyse temporal patterns of spiking, we took one trial (usually quently shown to a lesser extent for RSC HD neurons the first) for each cell and used the ISI histogram (Supplementary (Cho and Sharp, 2001). It is determined here by taking the Figure 4) to determine the typical time between spikes (ISI peak) slope of the AHV/PFD relationship. Comparing PoS and RSC and spread of firing intervals (decay time to half-peak; Table 1). Lozano et al. 9 Figure 3. Modulation of HD cell firing by angular head velocity (AHV). (a) The heat plots show firing rate (colour; max = red) as a function of HD (y-axis) and AHV (x-axis) for a PoS and RSC neuron. Tuning curves collapsed across all AHVs for left and right head turns are shown to the left and right sides, respectively. Firing rate collapsed across HD as a function AHV is shown below each plot (means and standard error of the mean (SEM) of the shuffled control shown in blue). The left–right downwards slope evident in the RSC plot reveals anticipatory firing, in which the tuning curve shifts in the positive direction (right) for left head turns and in the other direction for right head turns. The shift increased linearly with AHV, revealing a constant time lag. (b) Overall, anticipatory firing differed between PoS and RSC (dotted line shows mean). 10 Brain and Neuroscience Advances individual co-recorded cells in a trial was averaged to yield an overall value. Overall cue control We first looked at whether there was overall cue control by the cue pair, irrespective of the visual stimuli feature, and whether this would differ for brain region or cue type. As described in section ‘Methods’ (Figure 2), the firing directions for each trial in a session were extracted and re-oriented relative to the session mean, so as to express them all in the same reference frame. Then, because visual inspection suggested that firing directions were sometimes distributed in a bipolar fashion, we doubled the angles modulo 360 and re-computed the firing directions; this has the effect of wrapping points at 180° around to 0/360°, and rendering a bipolar distribution unipo- lar. The angle-doubled data were compared against a control, shuffled dataset in which the PFD for each trial was randomly generated; this procedure was repeated until 1000 such pseudo-sessions had been obtained. Overall, the 2140 cells × trials (149 cells in multiple trials) comprised 221 cells × sessions, which collapsed (after averag- ing across cells) to 113 sessions, of which 41 were from PoS and 72 from RSC. Overall, the number of cells recorded in each cue condition were 48, 166 and 17 for HIGH, MOD and LOW cues, respectively. The shuffle control procedure yielded R-vector scores of 0.31 ± 0.01 for both single and doubled val- ues, so this was used as the threshold against which to statisti- cally evaluate cue control. The population of all firing directions before and after angle- doubling are shown in Figure 4(a). R-vector scores showed a marked increase after angle-doubling, from 0.56 to 0.74, indicating Figure 4. Overall cue control. The circular dot plots show the raw cell a degree of bipolarity in the original firing directions, therefore to data, expressed relative to session means (0°); the histograms below evaluate overall cue control as a function of brain area, the angle- these show the same data binned in 6° bins, linearised (centred on doubled values were used. Overall, the angle-doubled R-vector 0°) and expressed as a proportion of the total cell count in order to scores far exceeded the shuffled control threshold (t(112) = 23.00, enable easier visual comparison between the cue types. (a) Angle- p < 0.0001), indicating a high degree of cue-following. Average singled or doubled mean firing directions relative to the cue pair, for per-session R-vector scores for PoS were 0.73 ± 0.04 and for the individual trials. For the original angles, each cell’s preferred firing RSC were 0.79 ± 0.02; these values did not differ (two-tailed direction (PFD) was computed relative to Cue 1, and mean values t(111) = 1.42, p = 0.16), indicating no difference between brain computed for each session, each PFD was then realigned to this session areas in the overall level of cue control (Figure 4(b)). mean (0°). For the doubled angles, the PFD value relative to the cue was doubled, modulo 360°, so as to wrap values near 180° around to Cue discrimination zero and thus remove the bimodality. (b) The doubled-angled values compared between PoS and RSC for all trials. Statistics reported in the We next looked at cue discrimination as a function of cue type, first text were calculated using the session means rather than the set of using the original non-doubled angles. If the cells were confusing individual trials. cues, then the firing should be bipolar and the R-vector scores low. Indeed, firing directions (Figure 5) were more clustered for the more discriminable cues. R-vector scores were much higher for the neurons (Figure 3(b)), we found that, as in previous studies, HIGH cues (0.73 ± 0.05) than for the MOD cues (0.55 ± 0.03) or RSC neurons showed a greater ATI (47.91 ± 4.28 ms) than did the LOW cues (0.27 ± 0.08). Unbalanced one-way ANOVA found PoS neurons (14.37 ± 8.19 ms); these were significantly differ- this difference to be highly significant (F(2, 110) = 8.0, p = 0.0006), ent (one-tailed t(147) = 3.62, p = 0.0002). The PoS values did and post hoc testing (Tukey’s honest significant difference (HSD)) not differ from zero (t(74) = 1.75, p = 0.0836). found that the HIGH cues had significantly higher R-vector scores than both the MOD cues (t(101) = 2.18, p = 0.03) and the LOW cues (t(23) = 6.58, p = 0.0001), while the MOD and LOW cues also Cue control differed (t(96) = 2.90, p = 0.005). The next set of analyses investigated ensemble behaviour and In order to determine whether the differences might be due aimed to determine the extent to which the cells were con- to differential confusion between cues, we repeated the analysis trolled by the visual cues: for these analyses, behaviour of the using the angle-doubled values to remove bipolarity. R-vector Lozano et al. 11 Figure 5. Comparison of responding to the three categories of visual cues. Dot plots and histograms are generated as for Figure 4. (a) Examples of eight recording trials from three HD cells, one from each cue condition, showing the relationship of firing direction (polar plot) to the index cue (black dot). Peak firing in Hz is shown below each plot. The first two cells are from RSC, the last from PoS. The cell in the HIGH discriminability condition showed much less variability, as was typical. (b) and (c) Population data. PoS and RSC data have been combined. (b) The original angles; (c) the doubled angles, showing increased dispersion for the highly discriminable cues (HIGH) but decreased dispersion for the MOD and LOW cues, indicating that their original firing directions had a degree of bipolarity. See text for statistical comparisons. 12 Brain and Neuroscience Advances differences in resulting R-vector score (F(2, 85) = 0.90, p = 0.39) so the data were analysed together. Basic single- angled R-vector scores for the two brain areas did not differ (PoS = 0.57 ± 0.05, RSC = 0.55 ± 0.04, t(85) = 0.23, p = 0.82). Similarly, although the doubled-angle scores significantly improved overall, from 0.55 ± 0.03 to 0.79 ± 0.02 (one-tailed −10 paired t(87) = 7.13, p = 1.37 × 10 ), the doubled-angle Rayleigh scores were not different (t(85) = 0.33, p = 0.74). Thus, there was no difference in the degree of cue control ver- sus cue confusion between brain areas. Comparison of visual and non-visual contributions to cue discrimination For five recording sessions (48 trials) from one rat (the sham LGN-lesioned one), a test was made to see whether the discrimi- nation was purely based on visual pattern, or whether there might have been a contribution from the olfactory/tactile com- ponents of the cue cards. This was done using the V-H and T-B cue cards, which could be reversed in visual identity (V to H, or T to B, and vice versa) by rotating them around their centre points (90° to make V become H, and 180° to make T become B). Data were then analysed with firing aligned in the visual-cue reference frame (where Cue 1 was the V or T cue, irrespective of physical card identity) or in the physical reference frame (where Cue 1 is the same card across trials, regardless of its orientation). The alignment of PFDs is shown in Figure 7. In both reference frames there was a degree of bipolarity, indicating that the PFD sometimes followed the secondary cue in a given reference frame, but cue control was stronger in the visual reference frame (R-vector of 0.54) than in the physical one (R = 0.30). When R-vectors were compared across sessions, there was a decline in the score when the visual-frame data were compared with the physical-frame data (visual = 0.54 ± 0.05; physical = 0.31 ± 0.08). This difference only just reached statistical significance Figure 6. Comparison of responding to the MOD cues. Preferred firing (t(4) = 2.2, one-tailed p = 0.046), but the improvement from sin- directions were derived and plotted, together with R-vector score and gle to double-angle R-scores hints at a contribution from both normalised histograms, as for Figure 5. There was an improvement sensory modalities, with visual being slightly stronger. in score with the doubled-angle values, indicating a degree of cue confusion, but no difference in cue control between PoS and RSC (see RSC and PoS HD cells produced text for details). unipolar tuning curves even when the landmarks were identical scores for the HIGH cues were 0.62 ± 0.06, for the MOD cues were 0.79 ± 0.04 and for the LOW cues were 0.77 ± 0.23. ANOVA showed that scores for the HIGH cues were now sig- No evidence of bipolar tuning curves was seen by visual inspec- nificantly lower than for the MOD cues (t(101) = 3.01, p = 0.003) tion. However, to check quantitatively, an autocorrelation was and not different for the LOW cues (t(23) = 1.63, p = 0.12); nor run on the data (see section ‘Methods’) in order to determine did the MOD and LOW cues differ (t(96) = 0.36, p = 0.72). whether there was a secondary peak at 180° due to periodic Thus, the superior single-angle R-vector scores of the HIGH reversal of the tuning curve arising from cue confusion. Session cues can be attributed to the more bipolar distribution of the data were combined into a single dataset for a given cell, and the MOD and LOW cues. autocorrelation peaks extracted. Identical-cue trials were ana- To compare between brain regions, we then looked in more lysed separately because there were too few of them: a two-way detail at responses to the MOD cues, as these were the ones ANOVA of the remaining conditions found no significant differ- that were in principle discriminable but in practice evoked ences, either as a function of cue type (F(4, 48 = 0.26, p = 0.90), some confusion, suggesting maximal demand on landmark dis- autocorrelation peak (F(1, 12) = 0.59, p = 0.46) or interaction crimination processing. However, we found no difference. The (F(4, 48 = 0.11, p = 0.36). A paired t-test on the identical-cue con- overall single- and doubled-angle data for all the MOD trials dition, where bipolarity might have been strongest if it occurred, are shown in Figure 6, as a function of brain area. Comparison was also non-significant (t(12) = 1.54, p = 0.07). Thus, there was of the three cue types (R-L, T-B and V-H) yielded no no hint of bipolarity in the tuning curves within a single trial. Lozano et al. 13 Figure 7. Reversal of firing provoked by visual change alone. Data from a rat in which cues were changed in visual identity independently of their physical identity, by rotating the cue cards so that a vertical bar became horizontal or a top bar became bottom, etc. When activity was aligned relative to the visual cue, activity was clustered, but slightly bipolar. When aligned to the physical cue identity, firing was more dispersed. When angles were doubled to remove bipolarity (which has the same final effect for both reference frames), the data were highly clustered around zero, indicating overall good control by the cue pair. See text for session statistics. cues (F(2, 228) = 0.2, p = 0.83). The y-intercept measure showed Stabilisation of PFD across a trial a barely significant difference (F(2, 228) = 3,1, p = 0.048), but this The time course of the establishment of firing direction was com- went in the opposite direction from expected (highly discrimina- puted using a stabilisation analysis (see section ‘Methods’), look- ble cues had a higher y-intercept, at 21.14 s) and may have been ing at either percentage of total spikes emitted in each 30-s time due to the low n (n = 17) for the LOW cues. decile, or time taken to achieve each spike count decile (Figure 8). In summary, the stabilisation analysis revealed a slight ten- The percentage of spikes emitted in the first 30 s was 8.27 ± 0.13%, dency for a slower accumulation of spikes (in other words, slower which was significantly less than the expected 10% (t(230) = 13.11, development of firing), especially in the PFD, in the first 20–30 s p < 0.0001). The z-score for this decile was −0.78, while for the of each trial, but this did not depend on brain region or cue type. remaining deciles it did not deviate more than 0.44 from zero Overall, this analysis reveals that actually there was a propensity (Figure 8(a)). A corresponding observation emerged from the for firing to become established very soon after entry into the y-intercept analysis. The plots in Figure 8(b) are highly linear, with environment, even for cues that were harder to discriminate. R values close to 1 for most sessions (0.9979 ± 0.0002 for the 231 cells × sessions overall). However, the regression lines did not go through zero: the y-intercepts averaged 18.22 +/- 1.05 s, which was Discussion significantly different from zero (t(230) = 21.44, p < 0.0001). Thus, This study compared the activity of PoS and RSC HD cells in a the time taken to accumulate the first decile’s worth of spikes in the landmark discrimination setting, in which two opposing land- PFD range was almost 20 s longer than would be expected based marks had similar overall shape and/or contrast/luminance but on the subsequent deciles. In order to determine whether this effect differed in their fine-grained visual structure. The study was was due to dispersion of firing around multiple directions or just to motivated by previous findings that lesions to the PoS or RSC a general decline in spiking, we repeated the analysis using all impair both directional tuning stability and landmark control of spikes, not just the ones in the PFD range; a similar, albeit slightly HD cells in other parts of the circuit, suggesting a role for these attenuated effect was evident (Figure 8), suggesting a general structures in directional landmark processing. The aims were reduction of activity, rather than spatial dispersion, in the first 20– twofold: (a) to see whether the two regions differed in their 30 s of environment exploration. We looked at movement corre- responsiveness, which might provide clues to their function, and lates including distribution of HD with respect to the PFD, linear (b) to see whether HD cells could distinguish the landmarks by velocity and AHV, and did not find any difference between the first maintaining a consistent directional relationship to the array, and decile and subsequent ones that could explain this observation (not if so, to determine how quickly this discrimination occurred, and shown). This slight reduction in spiking suggests that on entry into whether there were regional differences. These two considera- an environment there is a period of time during which the system tions are discussed in turn. is acquiring the information needed to drive the cells to firing threshold. We compared these measures between PoS and RSC but Firing properties of PoS versus RSC HD found no differences. The total percentage of spikes accumulated neurons in the first 30 s of the trial did not differ between PoS and RSC The basic firing properties we saw were similar to those reported (t(229) = 1.86, p = 0.06) and the y-intercept measure was also not previously, with the exception that the peak firing rates we observed different (t(229) = 1.87, p = 0.06). We finally looked at the two in PoS were considerably lower than those that have been reported measures as a function of cue type. The spike count percentages previously: ~8 Hz in our study compared with ~35 Hz reported by in the first 30 s did not differ between HIGH, MOD and LOW 14 Brain and Neuroscience Advances Figure 8. Stabilisation of the preferred firing direction (PFD) on environment entry. Data from a 40° range surrounding the PFD were divided into deciles of either time (30 s bins) or spikes (10% total). (a) Top: Spike count as a function of time decile revealed a reduced spike count in the first 30 s (circled). Bottom: z-score analysis of those data revealed a larger negative score for the first decile relative to the others, present for both spikes in the PFD range and to a lesser extent in all the spikes. See Results for statistical tests. (b) Top: The spikes from each 300 s trial were divided into 10 deciles, and the time to reach each decile plotted. Each fine grey line represents the plot from one cell over one session (averaged across trials); the solid black line is the mean across all cells/sessions and the dotted line extrapolates to the y-axis. Bottom: the y-intercepts, which should be at zero, were clustered around a mean (black line) of 18 s for the within-PFD spikes, indicating that it took extra time to reach the first decile relative to subsequent deciles. As with the spike count data, the effect was attenuated but still present when all spikes were considered. Taube et al. (1990b) and Taube and Muller (1998) and 19 Hz for The observation that RSC neurons anticipate upcoming HD’s, Sharp (1996). The reason for the large discrepancies between stud- in our case by 48 ms, is similar to that published previously. ies might be due to the recording methods used, and the possible Anticipatory firing was first reported by Blair and Sharp (1995) difficulty of the single-wire recording methods of earlier studies to for anterior thalamic HD neurons, which were found to predict identify low-firing-rate neurons. In support of this, a more recent future HD by around 37 ms; this contrasted with PoS which did study using tetrodes, as in the present study, found a rate of 3 Hz in not show anticipation. Taube and Muller (1998) reported a value PoS HD cells (Brandon et al., 2012). We also averaged across sev- of around 23 ms for anterior thalamus, and −7 ms (firing lagging eral trials, which may have brought the rates towards the mean. HD) for PoS HD neurons. In a later study, RSC HD neurons were Relatively few differences in firing properties were seen in HD also found to show anticipation, of around 25 ms (Cho and Sharp, cells from the two brain regions, but those that we observed repli- 2001). In this study, we also saw no anticipation in PoS neurons. cated previous observations: (a) firing rates were higher in RSC; The implication is that RSC firing very slightly precedes PoS fir- (b) tuning curves were slightly broader and (c) RSC firing antici- ing; it may be, therefore, that the route of movement-related pated HD by about 48 ms. The observation of higher firing rates in information flow is from RSC to PoS. This is consistent with the RSC is consistent with a report by Sharp (2005), who found similar observation that granular RSC neurons project to layer III PoS but smaller differences in firing rate between these regions. We (Kononenko and Witter, 2012), which is where many HD cells in also found that firing rate was weakly (albeit significantly) corre- this region are found (Boccara et al., 2010; Preston-Ferrer et al., lated with running speed, with no differences between PoS and 2016), the remainder being in the deep layers (Boccara et al., RSC. However, PoS was slightly more influenced by AHV 2010; Preston-Ferrer et al., 2016; Ranck, 1984; Taube et al., although modulation was low for both regions. 1990b). Anticipation may reflect vestibular or motor efference Lozano et al. 15 copy signals, providing advance warning of upcoming head as proposed both by attractor network models (Knierim and directions which can then be combined with descending land- Zhang, 2012) and following population recordings of HD cells mark-related sensory inputs. A study of ADN neurons found that (Peyrache et al., 2015; Seelig and Jayaraman, 2015). anticipatory firing also occurred, and indeed more prominently, Having established cue control by the landmark pair, we in association with passive head movements (Bassett et al., then looked to see whether the cues within the pair could be 2005), suggesting a stronger vestibular contribution (Van der further discriminated. We found that discrimination of the Meer et al., 2007). Clark et al. (2010) found that neurotoxic RSC non-identical cues was highest for the high-contrast HIGH lesions increase anticipatory firing in the ADN, possibly due to (black/white) cue pair but was also well above control levels removal of an environment-anchoring stimulus and consequent for the patterned, moderately discriminable (MOD) cues too. overweighting of the vestibular AHV inputs. The precise contri- For one rat, we dissociated visual and olfactory characteristics butions of the various self-motion signals, including how they of the cue cards and found a clear contribution from vision. may be differentially weighted in different species and/or under These data show that these cortical HD regions, which are one different conditions, remains to be determined (Cullen, 2014). synapse away from the visual cortex (Vogt and Miller, 1983; Because of the connections of both regions with the hip- Van Groen and Wyss, 1992), can access information about pocampal formation, and because some theoretical models of HD internal structure of landmarks – furthermore, discrimination cell processing require spatial inputs (Bicanski and Burgess, occurs almost immediately upon entry into the environment, 2016), we looked for spatial correlates of firing in the form of and remains stable thereafter. place fields, or at least of reliable spatial heterogeneity. Both When we looked at the period of time immediately after envi- brain regions yielded some cells showing weak spatial encoding, ronment entry we found a gradual increase in firing rate for the but in general spatial modulation was low. For PoS, this observa- first few seconds, reflected in a longer time to accumulate the first tion is consistent with previous studies from this area in which decile’s worth of spikes relative to subsequent deciles. Thus, it HD cells typically have neither place nor theta-frequency modu- seems that landmarks do not purely cause directional clustering of lation (Sharp, 1996). However, Cacucci et al. (2004) and Boccara a randomly distributed base level of activity, but add drive to the et al. (2010) did report conjunctive encoding in regions of pre- system over and above baseline. Two previous studies have subiculum very close to PoS – it is thus likely that there are looked at the dynamics of cue control and have found that regional differences in the occurrence of the difference cell types. although it takes a significant amount of time – minutes – for a Overall, however, it seems unlikely that spatial inputs provide a novel cue to gain control of HD cells (Goodridge et al., 1998), strong input to the HD cell signal directly (although the possibil- cells respond to rotation of a familiar cue within a few tens of mil- ity exists that spatial inputs gate landmark inputs and thus exert liseconds (Zugaro et al., 2003). Our situation is different because their influence in a more subtle manner). although the landmarks were generally familiar, measurements were made within the first few seconds of entry into the environ- ment, and also the cues needed to be discriminated. This could be Landmark discrimination due to the slow onset of intrinsic network stabilisation, or it may One purpose of this study was to examine possible differences in be that a number of cognitive factors come into play that influence landmark processing by the two regions. Cells from both regions HD cell firing in these first few seconds – the animal may need in all the different cue conditions had unipolar tuning curves that time to adjust to the situation and start attending to spatial cues, were clustered around a single direction relative to the cue pair, there may be a lag while the cues are identified and the discrimi- indicating that they could distinguish the cues from the back- nation is made and so on. It would be interesting to look at firing ground and reliably orient by them. This was true even for the rate if the animal were placed into the arena without the cues, to visually identical cues; there was no hint that within a trial, the try and tease apart these possibilities. cells flipped their firing from one direction to its polar opposite. The rapidity with which landmark discrimination was There are two possible reasons for this – one is that the rats could established, in the absence of explicit training, suggests that still distinguish the cues by non-visual means (e.g. perhaps by this two-cue HD cell paradigm may be a good experimental olfaction) and did in fact fire in a unipolar fashion with respect to method with which to test visual discrimination capabilities of the cues. This seems unlikely since there was confusion between rats: for example, following lesions to visual brain areas. cues even for the ones that were moderately discriminable; also, Because explicit training may bias animals to focus on single for the one rat in which the cues were identified by the experi- features of multidimensional stimuli (Jeffery, 2007), sponta- menter and visual and physical identities switched between trials, neous recognition is preferable, but such processes may be the firing directions were less well predicted by physical identity. hard to detect in behavioural output. Cell activity on the other The other, more likely reason is that firing was stabilised within a hand is directly observable and easily measured. The fact that trial by self-motion signals. In other words, having initially HD cells rapidly detect and discriminate complex stimuli guessed at one of the two possible orientations of the cue pair on means that they may serve as a useful read-out for sensory entry into the environment, the cells were then anchored to that perception occurring within seconds or even faster. orientation with the help of self-motion cues (since the rat knew, Overall, therefore, we found relatively few differences from these cues, that it had not suddenly reversed direction). between PoS and RSC in their cue responsiveness. The similarity Henceforth, the system could use the ambiguous landmark pair in responding can be explained either by similar computations together with the self-motion cues to generate a stable directional being performed on the incoming visual signals by the two regions signal, whereby the landmarks would prevent the signal drifting, or else by rapid communication between regions. It seems a priori and the self-motion cues would prevent it from flipping. This unlikely that the regions perform the same computations, but to proposition is consistent with the operation of attractor dynamics, address this question in more detail it will be necessary to isolate 16 Brain and Neuroscience Advances Cacucci F, Lever C, Wills TJ, et al. (2004) Theta-modulated place-by- the regions, either by lesioning each in turn or by targeted opto- or direction cells in the hippocampal formation in the rat. Journal of chemogenetic interventions that functionally disconnect them. Neuroscience 24(38): 8265–8277. This study additionally introduces a new spontaneous visual dis- Calton JL, Stackman RW, Goodridge JP, et al. (2003) Hippocampal crimination method for probing visual processing by the spatial place cell instability after lesions of the head direction cell network. system and reveals a basic capacity for the two cortical HD cell Journal of Neuroscience 23(30): 9719–9731. areas to rapidly detect and discriminate landmarks. These experi- Calton JL, Turner CS, Cyrenne D-LM, Lee BR, Taube JS (2008) Land- ments open the door to investigations of transformation of the mark control and updating of self-movement cues are largely main- visual pathway following lesions or selective inactivation of tained in head direction cells after lesions of the posterior parietal inputs to these areas. Our prediction is that even when the object cortex. Behavioral Neuroscience 122(4): 827–840. processing regions of the brain, such as perirhinal cortex, are Cho J and Sharp PE (2001) Head direction, place, and movement corre- lates for cells in the rat retrosplenial cortex. Behavioral Neuroscience removed, HD neurons will still be able, using basic visual inputs 115(1): 3–25. from primary visual areas, to discriminate similar landmarks and Clark BJ, Bassett JP, Wang SS, et al. (2010) Impaired head direction cell use them for orientation. representation in the anterodorsal thalamus after lesions of the retro- splenial cortex. Journal of Neuroscience 30(15): 5289–5302. Acknowledgements Cullen KE (2014) The neural encoding of self-generated and externally applied movement: Implications for the perception of self-motion The authors are grateful to Miguel Valencia for advice on analysis and and spatial memory. Frontiers Integrative Neuroscience 7: 108. Josh Bassett for useful comments on the article. Golob EJ, Stackman RW, Wong AC, et al. (2001) On the behavioral significance of head direction cells: Neural and behavioral dynam- ics during spatial memory tasks. Behavioral Neuroscience 115(2): Data sharing 285–304. Data supporting the findings of this study are available on request from Goodridge JP and Taube JS (1997) Interaction between the postsubicu- the corresponding author. lum and anterior thalamus in the generation of head direction cell activity. Journal of Neuroscience 17(23): 9315–9330. Goodridge JP, Dudchenko P, Worboys K, et al. (1998) Cue control and Declaration of conflicting interests head direction cells. Behavioral Neuroscience 112(4): 749–761. KJ is a non-shareholding director of Axona Ltd. Jacob PY, Casali G, Spieser L, et al. (2016) An independent, landmark- dominated head direction signal in dysgranular retrosplenial cortex. Nature Neuroscience 20(2): 173–175. Funding Jankowski MM, Islam MN, Wright NF, et al. (2014) Nucleus reuniens of The work was supported by funding from the UK Medical Research the thalamus contains head direction cells. eLife 3: e03075. Council (G1100669) and the Wellcome Trust (WT103896AIA) to KJ Jeffery KJ (2007) Do discrimination tasks discourage multi-dimensional and a studentship from the Mexican Consejo Nacional de Ciencia y stimulus processing? Evidence from a cross-modal object discrimi- Tecnologia (3132554) to YL. nation in rats. Behavioural Brain Research 183(2): 213–221. Kadir SN, Goodman DFM and Harris KD (2014) High-dimensional clus- ter analysis with the masked EM algorithm. Neural Computation References 26(11): 2379–2394. Abeles M (1982) Quantification, smoothing, and confidence limits for Knierim JJ and Zhang K (2012) Attractor dynamics of spatially corre- single-units’ histograms. Journal of Neuroscience Methods 5(4): lated neural activity in the limbic system. Annual Review of Neuro- 317–325. science 35: 267–285. Bassett JP and Taube JS (2001) Neural correlates for angular head veloc- Knight R, Hayman R, Ginzberg LL, et al. (2011) Geometric cues influ- ity in the rat dorsal tegmental nucleus. Journal of Neuroscience ence head direction cells only weakly in non-disoriented rats. Jour- 21(15): 5740–5751. nal of Neuroscience 31(44): 15681–15692. Bassett JP, Zugaro B, Muir GM, et al. (2005) Passive movements of the Knight R, Piette CE, Page H, et al. (2014) Weighted cue integration in head do not abolish anticipatory firing properties of head direction the rodent head direction system. Philosophical Transaction of the cells. Journal of Neurophysiology 93(3): 1304–1316. Royal Society of London, Series B: Biological Science 369(1635): Berens P (2009) CircStat: A MATLAB toolbox for circular statistics. Journal of Statistical Software 31(10): 1–21. Kononenko NL and Witter MP (2012) Presubiculum layer III conveys Bicanski A and Burgess N (2016) Environmental anchoring of head retrosplenial input to the medial entorhinal cortex. Hippocampus direction in a computational model of retrosplenial cortex. Journal 22(4): 881–895. of Neuroscience 36(46): 11601–11618. Paxinos G and Watson C (2007) The Rat Brain in Stereotaxic Coordi- Blair HT and Sharp PE (1995) Anticipatory head direction signals in nates (6th edn). Cambridge, MA: Elsevier Academic Press, pp. anterior thalamus: Evidence for a thalamocortical circuit that inte- 547–612. grates angular head motion to compute head direction. Journal of Peyrache A, Lacroix MM, Petersen PC, et al. (2015) Internally organized Neuroscience 15(9): 6260–6270. mechanisms of the head direction sense. Nature Neuroscience 18(4): Blair HT, Lipscomb BW and Sharp PE (1997) Anticipatory time intervals 569–575. of head-direction cells in the anterior thalamus of the rat: Implica- Peyrache A, Roux L, Schieferstein N, et al. (2016) Transformation of tions for path integration in the head-direction circuit. Journal of head-direction signal into spatial code. bioRxiv. Epub ahead of print Neurophysiology 78(1): 145–159. 20 September. DOI: 10.1101/075986. Boccara CN, Sargolini F, Thoresen VH, et al. (2010) Grid cells in pre- Preston-Ferrer P, Coletta S, Frey M, et al. (2016) Anatomical organization and parasubiculum. Nature Neuroscience 13(8): 987–994. of presubicular head-direction circuits. eLife 5: e14592. Brandon MP, Bogaard AR, Andrews CM, et al. (2012) Head direc- Ranck JB (1984) Head-direction cells in the deep layers of the dor- tion cells in the postsubiculum do not show replay of prior waking sal presubiculum in freely moving rats. Society for Neuroscience sequences during sleep. Hippocampus 22(3): 604–618. 10: 599. Lozano et al. 17 Seelig JD and Jayaraman V (2015) Neural dynamics for landmark ori- Taube JS, Muller RU and Ranck JB Jr (1990a) Head-direction cells entation and angular path integration. Nature 521(7551): 186–191. recorded from the postsubiculum in freely moving rats I: Descrip- Sharp PE (1996) Multiple spatial/behavioral correlates for cells in the rat tion and quantitative analysis. Journal of Neuroscience 10(2): postsubiculum: Multiple regression analysis and comparison to other 420–435. hippocampal areas. Cerebral Cortex 6(2): 238–259. Taube JS, Muller RU and Ranck JB Jr (1990b) Head-direction cells Sharp PE (2005) Regional distribution and variation in the firing proper- recorded from the postsubiculum in freely moving rats II: Effects ties of head direction cells. In: Wiener SI and Taube JS (eds) Head of environmental manipulations. Journal of Neuroscience 10(2): Direction Cells and the Neural Mechanisms of Spatial Orientation. 436–447. Cambridge, MA: MIT Press, pp. 3–15. Van der Meer M, Knierim JJ, Yoganarasimha D, et al. (2007) Anticipa- Skaggs WE, Gothard KM and Markus EJ (1993) An information-the- tion in the rodent head direction system can be explained by an inter- oretic approach to deciphering the hippocampal code. In: Hanson action of head movements and vestibular firing properties. Journal SJ, Cowan JD and Glies CL (eds) Advances in Neural Information of Neurophysiology 98(4): 1883–1897. Process Systems 5. San Francisco, CA: Morgan Kaufmann Publish- Van Groen T and Wyss JM (1992) Connections of the retrosplenial dys- ers, pp. 1031–1037. granular cortex in the rat. Journal of Comparative Neurology 315(2): Taube JS (2007) The head direction signal: Origins and sensory-motor 200–216. integration. Annual Review of Neuroscience 30(1): 181–207. Vogt BA and Miller MW (1983) Cortical connections between rat cingu- Song P and Wang X-J (2005) Angular path integration by moving “hill of late cortex and visual, motor, and postsubicular cortices. Journal of activity”: A spiking neuron model without recurrent excitation of the Comparative Neurology 216(2): 192–210. head-direction system. Journal of Neuroscience 25(4): 1002–1014. Yoder RM, Clark BJ and Taube JS (2011) Origins of landmark encoding Taube JS and Burton HL (1995) Head direction cell activity monitored in in the brain. Trends in Neurosciences 34(11): 561–571. a novel environment and during a cue conflict situation. Journal of Yoder RM, Peck JR and Taube JS (2015) Visual landmark information Neurophysiology 74(5): 1953–1971. gains control of the head direction signal at the lateral mammillary Taube JS and Muller RU (1998) Comparisons of head direction cell nuclei. Journal of Neuroscience 35(4): 1354–1367. activity in the postsubiculum and anterior thalamus of freely moving Zugaro MB, Arleo A, Berthoz A, et al. (2003) Rapid spatial reorientation rats. Hippocampus 8(2): 87–108. and head direction cells. Journal of Neuroscience 23(8): 3478–3482. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Brain and Neuroscience Advances SAGE http://www.deepdyve.com/lp/sage/retrosplenial-and-postsubicular-head-direction-cells-compared-during-ZU634QvrfV

Loading next page...

References (99)

( JankowskiMMIslamMNWrightNFet al (2014) Nucleus reuniens of the thalamus contains head direction cells. eLife 3: e03075.)
JankowskiMMIslamMNWrightNFet al (2014) Nucleus reuniens of the thalamus contains head direction cells. eLife 3: e03075.
JankowskiMMIslamMNWrightNFet al (2014) Nucleus reuniens of the thalamus contains head direction cells. eLife 3: e03075., JankowskiMMIslamMNWrightNFet al (2014) Nucleus reuniens of the thalamus contains head direction cells. eLife 3: e03075.
( RanckJB (1984) Head-direction cells in the deep layers of the dorsal presubiculum in freely moving rats. Society for Neuroscience 10: 599.)
RanckJB (1984) Head-direction cells in the deep layers of the dorsal presubiculum in freely moving rats. Society for Neuroscience 10: 599.
RanckJB (1984) Head-direction cells in the deep layers of the dorsal presubiculum in freely moving rats. Society for Neuroscience 10: 599., RanckJB (1984) Head-direction cells in the deep layers of the dorsal presubiculum in freely moving rats. Society for Neuroscience 10: 599.
J. Calton, R. Stackman, J. Goodridge, W. Archey, P. Dudchenko, J. Taube
Behavioral/systems/cognitive Hippocampal Place Cell Instability after Lesions of the Head Direction Cell Network
( GoodridgeJPDudchenkoPWorboysKet al (1998) Cue control and head direction cells. Behavioral Neuroscience 112(4): 749–761.9733184)
GoodridgeJPDudchenkoPWorboysKet al (1998) Cue control and head direction cells. Behavioral Neuroscience 112(4): 749–761.9733184
GoodridgeJPDudchenkoPWorboysKet al (1998) Cue control and head direction cells. Behavioral Neuroscience 112(4): 749–761.9733184, GoodridgeJPDudchenkoPWorboysKet al (1998) Cue control and head direction cells. Behavioral Neuroscience 112(4): 749–761.9733184
( BassettJPTaubeJS (2001) Neural correlates for angular head velocity in the rat dorsal tegmental nucleus. Journal of Neuroscience 21(15): 5740–5751.11466446)
BassettJPTaubeJS (2001) Neural correlates for angular head velocity in the rat dorsal tegmental nucleus. Journal of Neuroscience 21(15): 5740–5751.11466446
BassettJPTaubeJS (2001) Neural correlates for angular head velocity in the rat dorsal tegmental nucleus. Journal of Neuroscience 21(15): 5740–5751.11466446, BassettJPTaubeJS (2001) Neural correlates for angular head velocity in the rat dorsal tegmental nucleus. Journal of Neuroscience 21(15): 5740–5751.11466446
( BlairHTSharpPE (1995) Anticipatory head direction signals in anterior thalamus: Evidence for a thalamocortical circuit that integrates angular head motion to compute head direction. Journal of Neuroscience 15(9): 6260–6270.7666208)
BlairHTSharpPE (1995) Anticipatory head direction signals in anterior thalamus: Evidence for a thalamocortical circuit that integrates angular head motion to compute head direction. Journal of Neuroscience 15(9): 6260–6270.7666208
BlairHTSharpPE (1995) Anticipatory head direction signals in anterior thalamus: Evidence for a thalamocortical circuit that integrates angular head motion to compute head direction. Journal of Neuroscience 15(9): 6260–6270.7666208, BlairHTSharpPE (1995) Anticipatory head direction signals in anterior thalamus: Evidence for a thalamocortical circuit that integrates angular head motion to compute head direction. Journal of Neuroscience 15(9): 6260–6270.7666208
( GoodridgeJPTaubeJS (1997) Interaction between the postsubiculum and anterior thalamus in the generation of head direction cell activity. Journal of Neuroscience 17(23): 9315–9330.9364077)
GoodridgeJPTaubeJS (1997) Interaction between the postsubiculum and anterior thalamus in the generation of head direction cell activity. Journal of Neuroscience 17(23): 9315–9330.9364077
GoodridgeJPTaubeJS (1997) Interaction between the postsubiculum and anterior thalamus in the generation of head direction cell activity. Journal of Neuroscience 17(23): 9315–9330.9364077, GoodridgeJPTaubeJS (1997) Interaction between the postsubiculum and anterior thalamus in the generation of head direction cell activity. Journal of Neuroscience 17(23): 9315–9330.9364077
R. Yoder, James Peck, J. Taube (2015)
Visual Landmark Information Gains Control of the Head Direction Signal at the Lateral Mammillary Nuclei
The Journal of Neuroscience, 35
( VogtBAMillerMW (1983) Cortical connections between rat cingulate cortex and visual, motor, and postsubicular cortices. Journal of Comparative Neurology 216(2): 192–210.6863602)
VogtBAMillerMW (1983) Cortical connections between rat cingulate cortex and visual, motor, and postsubicular cortices. Journal of Comparative Neurology 216(2): 192–210.6863602
VogtBAMillerMW (1983) Cortical connections between rat cingulate cortex and visual, motor, and postsubicular cortices. Journal of Comparative Neurology 216(2): 192–210.6863602, VogtBAMillerMW (1983) Cortical connections between rat cingulate cortex and visual, motor, and postsubicular cortices. Journal of Comparative Neurology 216(2): 192–210.6863602
Jeiwon Cho, P. Sharp (2001)
Head direction, place, and movement correlates for cells in the rat retrosplenial cortex.
Behavioral neuroscience, 115 1
( YoderRMPeckJRTaubeJS (2015) Visual landmark information gains control of the head direction signal at the lateral mammillary nuclei. Journal of Neuroscience 35(4): 1354–1367.25632114)
YoderRMPeckJRTaubeJS (2015) Visual landmark information gains control of the head direction signal at the lateral mammillary nuclei. Journal of Neuroscience 35(4): 1354–1367.25632114
YoderRMPeckJRTaubeJS (2015) Visual landmark information gains control of the head direction signal at the lateral mammillary nuclei. Journal of Neuroscience 35(4): 1354–1367.25632114, YoderRMPeckJRTaubeJS (2015) Visual landmark information gains control of the head direction signal at the lateral mammillary nuclei. Journal of Neuroscience 35(4): 1354–1367.25632114
( Van der MeerMKnierimJJYoganarasimhaDet al (2007) Anticipation in the rodent head direction system can be explained by an interaction of head movements and vestibular firing properties. Journal of Neurophysiology 98(4): 1883–1897.17596421)
Van der MeerMKnierimJJYoganarasimhaDet al (2007) Anticipation in the rodent head direction system can be explained by an interaction of head movements and vestibular firing properties. Journal of Neurophysiology 98(4): 1883–1897.17596421
Van der MeerMKnierimJJYoganarasimhaDet al (2007) Anticipation in the rodent head direction system can be explained by an interaction of head movements and vestibular firing properties. Journal of Neurophysiology 98(4): 1883–1897.17596421, Van der MeerMKnierimJJYoganarasimhaDet al (2007) Anticipation in the rodent head direction system can be explained by an interaction of head movements and vestibular firing properties. Journal of Neurophysiology 98(4): 1883–1897.17596421
( KnierimJJZhangK (2012) Attractor dynamics of spatially correlated neural activity in the limbic system. Annual Review of Neuroscience 35: 267–285.)
KnierimJJZhangK (2012) Attractor dynamics of spatially correlated neural activity in the limbic system. Annual Review of Neuroscience 35: 267–285.
KnierimJJZhangK (2012) Attractor dynamics of spatially correlated neural activity in the limbic system. Annual Review of Neuroscience 35: 267–285., KnierimJJZhangK (2012) Attractor dynamics of spatially correlated neural activity in the limbic system. Annual Review of Neuroscience 35: 267–285.
( ChoJSharpPE (2001) Head direction, place, and movement correlates for cells in the rat retrosplenial cortex. Behavioral Neuroscience 115(1): 3–25.11256450)
ChoJSharpPE (2001) Head direction, place, and movement correlates for cells in the rat retrosplenial cortex. Behavioral Neuroscience 115(1): 3–25.11256450
ChoJSharpPE (2001) Head direction, place, and movement correlates for cells in the rat retrosplenial cortex. Behavioral Neuroscience 115(1): 3–25.11256450, ChoJSharpPE (2001) Head direction, place, and movement correlates for cells in the rat retrosplenial cortex. Behavioral Neuroscience 115(1): 3–25.11256450
Pierre-Yves Jacob, G. Casali, L. Spieser, H. Page, D. Overington, K. Jeffery (2016)
An independent, landmark-dominated head direction signal in dysgranular retrosplenial cortex
Nature neuroscience, 20
( BoccaraCNSargoliniFThoresenVHet al (2010) Grid cells in pre- and parasubiculum. Nature Neuroscience 13(8): 987–994.20657591)
BoccaraCNSargoliniFThoresenVHet al (2010) Grid cells in pre- and parasubiculum. Nature Neuroscience 13(8): 987–994.20657591
BoccaraCNSargoliniFThoresenVHet al (2010) Grid cells in pre- and parasubiculum. Nature Neuroscience 13(8): 987–994.20657591, BoccaraCNSargoliniFThoresenVHet al (2010) Grid cells in pre- and parasubiculum. Nature Neuroscience 13(8): 987–994.20657591
( PaxinosGWatsonC (2007) The Rat Brain in Stereotaxic Coordinates (6th edn). Cambridge, MA: Elsevier Academic Press, pp. 547–612.)
PaxinosGWatsonC (2007) The Rat Brain in Stereotaxic Coordinates (6th edn). Cambridge, MA: Elsevier Academic Press, pp. 547–612.
PaxinosGWatsonC (2007) The Rat Brain in Stereotaxic Coordinates (6th edn). Cambridge, MA: Elsevier Academic Press, pp. 547–612., PaxinosGWatsonC (2007) The Rat Brain in Stereotaxic Coordinates (6th edn). Cambridge, MA: Elsevier Academic Press, pp. 547–612.
Mark Brandon, A. Bogaard, Chris Andrews, Michael Hasselmo (2012)
Head direction cells in the postsubiculum do not show replay of prior waking sequences during sleep
Hippocampus, 22
J. Bassett, Michaël Zugaro, G. Muir, E. Golob, R. Muller, J. Taube (2005)
Passive movements of the head do not abolish anticipatory firing properties of head direction cells.
Journal of neurophysiology, 93 3
( Van GroenTWyssJM (1992) Connections of the retrosplenial dysgranular cortex in the rat. Journal of Comparative Neurology 315(2): 200–216.1545009)
Van GroenTWyssJM (1992) Connections of the retrosplenial dysgranular cortex in the rat. Journal of Comparative Neurology 315(2): 200–216.1545009
Van GroenTWyssJM (1992) Connections of the retrosplenial dysgranular cortex in the rat. Journal of Comparative Neurology 315(2): 200–216.1545009, Van GroenTWyssJM (1992) Connections of the retrosplenial dysgranular cortex in the rat. Journal of Comparative Neurology 315(2): 200–216.1545009
Hugh Blair, Brian Lipscomb, Patricia Sharp (1997)
Anticipatory time intervals of head-direction cells in the anterior thalamus of the rat: implications for path integration in the head-direction circuit.
Journal of neurophysiology, 78 1
T. Groen, J. Wyss (1992)
Connections of the retrosplenial dysgranular cortex in the rat
Journal of Comparative Neurology, 315
R. Yoder, B. Clark, J. Taube (2011)
Origins of landmark encoding in the brain
Trends in Neurosciences, 34
J. Bassett, J. Taube (2001)
Neural Correlates for Angular Head Velocity in the Rat Dorsal Tegmental Nucleus
The Journal of Neuroscience, 21
E. Golob, R. Stackman, Andrew Wong, J. Taube (2001)
On the behavioral significance of head direction cells: neural and behavioral dynamics during spatial memory tasks.
Behavioral neuroscience, 115 2
Shabnam Kadir, Dan Goodman, K. Harris (2013)
High-Dimensional Cluster Analysis with the Masked EM Algorithm
Neural Computation, 26
Regional distribution and variation in the firing properties of head direction cells
( SkaggsWEGothardKMMarkusEJ (1993) An information-theoretic approach to deciphering the hippocampal code. In: HansonSJCowanJDGliesCL (eds) Advances in Neural Information Process Systems 5. San Francisco, CA: Morgan Kaufmann Publishers, pp. 1031–1037.)
SkaggsWEGothardKMMarkusEJ (1993) An information-theoretic approach to deciphering the hippocampal code. In: HansonSJCowanJDGliesCL (eds) Advances in Neural Information Process Systems 5. San Francisco, CA: Morgan Kaufmann Publishers, pp. 1031–1037.
SkaggsWEGothardKMMarkusEJ (1993) An information-theoretic approach to deciphering the hippocampal code. In: HansonSJCowanJDGliesCL (eds) Advances in Neural Information Process Systems 5. San Francisco, CA: Morgan Kaufmann Publishers, pp. 1031–1037., SkaggsWEGothardKMMarkusEJ (1993) An information-theoretic approach to deciphering the hippocampal code. In: HansonSJCowanJDGliesCL (eds) Advances in Neural Information Process Systems 5. San Francisco, CA: Morgan Kaufmann Publishers, pp. 1031–1037.
Matthijs Meer, J. Knierim, D. Yoganarasimha, E. Wood, Mark Rossum (2007)
Anticipation in the rodent head direction system can be explained by an interaction of head movements and vestibular firing properties.
Journal of neurophysiology, 98 4
G. Paxinos, Charles Watson (1983)
The Rat Brain in Stereotaxic Coordinates
( JefferyKJ (2007) Do discrimination tasks discourage multi-dimensional stimulus processing? Evidence from a cross-modal object discrimination in rats. Behavioural Brain Research 183(2): 213–221.17692934)
JefferyKJ (2007) Do discrimination tasks discourage multi-dimensional stimulus processing? Evidence from a cross-modal object discrimination in rats. Behavioural Brain Research 183(2): 213–221.17692934
JefferyKJ (2007) Do discrimination tasks discourage multi-dimensional stimulus processing? Evidence from a cross-modal object discrimination in rats. Behavioural Brain Research 183(2): 213–221.17692934, JefferyKJ (2007) Do discrimination tasks discourage multi-dimensional stimulus processing? Evidence from a cross-modal object discrimination in rats. Behavioural Brain Research 183(2): 213–221.17692934
( KononenkoNLWitterMP (2012) Presubiculum layer III conveys retrosplenial input to the medial entorhinal cortex. Hippocampus 22(4): 881–895.21710546)
KononenkoNLWitterMP (2012) Presubiculum layer III conveys retrosplenial input to the medial entorhinal cortex. Hippocampus 22(4): 881–895.21710546
KononenkoNLWitterMP (2012) Presubiculum layer III conveys retrosplenial input to the medial entorhinal cortex. Hippocampus 22(4): 881–895.21710546, KononenkoNLWitterMP (2012) Presubiculum layer III conveys retrosplenial input to the medial entorhinal cortex. Hippocampus 22(4): 881–895.21710546
K. Jeffery (2007)
Do discrimination tasks discourage multi-dimensional stimulus processing? Evidence from a cross-modal object discrimination in rats
Behavioural Brain Research, 183
K. Cullen, Paul Smith, Patrizia Fattori, A. Brichta
Integrative Neuroscience Mini Review Article the Neural Encoding of Self-generated and Externally Applied Movement: Implications for the Perception of Self-motion and Spatial Memory Functionally Analogous Cells Types in the Vestibular Pathways of Monkey and Mouse
J. Taube, R. Muller, JB Ranck (1990)
Head-direction cells recorded from the postsubiculum in freely moving rats. II. Effects of environmental manipulations
, 10
( SharpPE (2005) Regional distribution and variation in the firing properties of head direction cells. In: WienerSITaubeJS (eds) Head Direction Cells and the Neural Mechanisms of Spatial Orientation. Cambridge, MA: MIT Press, pp. 3–15.)
SharpPE (2005) Regional distribution and variation in the firing properties of head direction cells. In: WienerSITaubeJS (eds) Head Direction Cells and the Neural Mechanisms of Spatial Orientation. Cambridge, MA: MIT Press, pp. 3–15.
SharpPE (2005) Regional distribution and variation in the firing properties of head direction cells. In: WienerSITaubeJS (eds) Head Direction Cells and the Neural Mechanisms of Spatial Orientation. Cambridge, MA: MIT Press, pp. 3–15., SharpPE (2005) Regional distribution and variation in the firing properties of head direction cells. In: WienerSITaubeJS (eds) Head Direction Cells and the Neural Mechanisms of Spatial Orientation. Cambridge, MA: MIT Press, pp. 3–15.
J. Knierim, Kechen Zhang (2012)
Attractor dynamics of spatially correlated neural activity in the limbic system.
Annual review of neuroscience, 35
J. Calton, Carol Turner, De-Laine Cyrenne, Brian Lee, J. Taube (2008)
Landmark control and updating of self-movement cues are largely maintained in head direction cells after lesions of the posterior parietal cortex.
Behavioral neuroscience, 122 4
( ZugaroMBArleoABerthozAet al (2003) Rapid spatial reorientation and head direction cells. Journal of Neuroscience 23(8): 3478–3482.12716956)
ZugaroMBArleoABerthozAet al (2003) Rapid spatial reorientation and head direction cells. Journal of Neuroscience 23(8): 3478–3482.12716956
ZugaroMBArleoABerthozAet al (2003) Rapid spatial reorientation and head direction cells. Journal of Neuroscience 23(8): 3478–3482.12716956, ZugaroMBArleoABerthozAet al (2003) Rapid spatial reorientation and head direction cells. Journal of Neuroscience 23(8): 3478–3482.12716956
( SeeligJDJayaramanV (2015) Neural dynamics for landmark orientation and angular path integration. Nature 521(7551): 186–191.25971509)
SeeligJDJayaramanV (2015) Neural dynamics for landmark orientation and angular path integration. Nature 521(7551): 186–191.25971509
SeeligJDJayaramanV (2015) Neural dynamics for landmark orientation and angular path integration. Nature 521(7551): 186–191.25971509, SeeligJDJayaramanV (2015) Neural dynamics for landmark orientation and angular path integration. Nature 521(7551): 186–191.25971509
Maciej Jankowski, Md Islam, N. Wright, S. Vann, J. Erichsen, J. Aggleton, S. O’Mara (2014)
Nucleus reuniens of the thalamus contains head direction cells
eLife, 3
( Preston-FerrerPColettaSFreyMet al (2016) Anatomical organization of presubicular head-direction circuits. eLife 5: e14592.27282390)
Preston-FerrerPColettaSFreyMet al (2016) Anatomical organization of presubicular head-direction circuits. eLife 5: e14592.27282390
Preston-FerrerPColettaSFreyMet al (2016) Anatomical organization of presubicular head-direction circuits. eLife 5: e14592.27282390, Preston-FerrerPColettaSFreyMet al (2016) Anatomical organization of presubicular head-direction circuits. eLife 5: e14592.27282390
( SharpPE (1996) Multiple spatial/behavioral correlates for cells in the rat postsubiculum: Multiple regression analysis and comparison to other hippocampal areas. Cerebral Cortex 6(2): 238–259.8670654)
SharpPE (1996) Multiple spatial/behavioral correlates for cells in the rat postsubiculum: Multiple regression analysis and comparison to other hippocampal areas. Cerebral Cortex 6(2): 238–259.8670654
SharpPE (1996) Multiple spatial/behavioral correlates for cells in the rat postsubiculum: Multiple regression analysis and comparison to other hippocampal areas. Cerebral Cortex 6(2): 238–259.8670654, SharpPE (1996) Multiple spatial/behavioral correlates for cells in the rat postsubiculum: Multiple regression analysis and comparison to other hippocampal areas. Cerebral Cortex 6(2): 238–259.8670654
( TaubeJSBurtonHL (1995) Head direction cell activity monitored in a novel environment and during a cue conflict situation. Journal of Neurophysiology 74(5): 1953–1971.8592189)
TaubeJSBurtonHL (1995) Head direction cell activity monitored in a novel environment and during a cue conflict situation. Journal of Neurophysiology 74(5): 1953–1971.8592189
TaubeJSBurtonHL (1995) Head direction cell activity monitored in a novel environment and during a cue conflict situation. Journal of Neurophysiology 74(5): 1953–1971.8592189, TaubeJSBurtonHL (1995) Head direction cell activity monitored in a novel environment and during a cue conflict situation. Journal of Neurophysiology 74(5): 1953–1971.8592189
( CacucciFLeverCWillsTJet al (2004) Theta-modulated place-by-direction cells in the hippocampal formation in the rat. Journal of Neuroscience 24(38): 8265–8277.15385610)
CacucciFLeverCWillsTJet al (2004) Theta-modulated place-by-direction cells in the hippocampal formation in the rat. Journal of Neuroscience 24(38): 8265–8277.15385610
CacucciFLeverCWillsTJet al (2004) Theta-modulated place-by-direction cells in the hippocampal formation in the rat. Journal of Neuroscience 24(38): 8265–8277.15385610, CacucciFLeverCWillsTJet al (2004) Theta-modulated place-by-direction cells in the hippocampal formation in the rat. Journal of Neuroscience 24(38): 8265–8277.15385610
( CullenKE (2014) The neural encoding of self-generated and externally applied movement: Implications for the perception of self-motion and spatial memory. Frontiers Integrative Neuroscience 7: 108.)
CullenKE (2014) The neural encoding of self-generated and externally applied movement: Implications for the perception of self-motion and spatial memory. Frontiers Integrative Neuroscience 7: 108.
CullenKE (2014) The neural encoding of self-generated and externally applied movement: Implications for the perception of self-motion and spatial memory. Frontiers Integrative Neuroscience 7: 108., CullenKE (2014) The neural encoding of self-generated and externally applied movement: Implications for the perception of self-motion and spatial memory. Frontiers Integrative Neuroscience 7: 108.
B. Vogt, Michael Miller (1983)
Cortical connections between rat cingulate cortex and visual, motor, and postsubicular cortices
Journal of Comparative Neurology, 216
P. Sharp (1996)
Multiple spatial/behavioral correlates for cells in the rat postsubiculum: multiple regression analysis and comparison to other hippocampal areas.
Cerebral cortex, 6 2
( KnightRHaymanRGinzbergLLet al (2011) Geometric cues influence head direction cells only weakly in non-disoriented rats. Journal of Neuroscience 31(44): 15681–15692.22049411)
KnightRHaymanRGinzbergLLet al (2011) Geometric cues influence head direction cells only weakly in non-disoriented rats. Journal of Neuroscience 31(44): 15681–15692.22049411
KnightRHaymanRGinzbergLLet al (2011) Geometric cues influence head direction cells only weakly in non-disoriented rats. Journal of Neuroscience 31(44): 15681–15692.22049411, KnightRHaymanRGinzbergLLet al (2011) Geometric cues influence head direction cells only weakly in non-disoriented rats. Journal of Neuroscience 31(44): 15681–15692.22049411
J. Taube, R. Muller, JB Ranck (1990)
Head-direction cells recorded from the postsubiculum in freely moving rats. I. Description and quantitative analysis
, 10
( BicanskiABurgessN (2016) Environmental anchoring of head direction in a computational model of retrosplenial cortex. Journal of Neuroscience 36(46): 11601–11618.27852770)
BicanskiABurgessN (2016) Environmental anchoring of head direction in a computational model of retrosplenial cortex. Journal of Neuroscience 36(46): 11601–11618.27852770
BicanskiABurgessN (2016) Environmental anchoring of head direction in a computational model of retrosplenial cortex. Journal of Neuroscience 36(46): 11601–11618.27852770, BicanskiABurgessN (2016) Environmental anchoring of head direction in a computational model of retrosplenial cortex. Journal of Neuroscience 36(46): 11601–11618.27852770
R. Knight, R. Hayman, Lin Ginzberg, K. Jeffery (2011)
Geometric Cues Influence Head Direction Cells Only Weakly in Nondisoriented Rats
The Journal of Neuroscience, 31
Charlotte Boccara, F. Sargolini, Veslemøy Thoresen, T. Solstad, M. Witter, E. Moser, M. Moser (2010)
Grid cells in pre- and parasubiculum
Nature Neuroscience, 13
( ClarkBJBassettJPWangSSet al (2010) Impaired head direction cell representation in the anterodorsal thalamus after lesions of the retrosplenial cortex. Journal of Neuroscience 30(15): 5289–5302.20392951)
ClarkBJBassettJPWangSSet al (2010) Impaired head direction cell representation in the anterodorsal thalamus after lesions of the retrosplenial cortex. Journal of Neuroscience 30(15): 5289–5302.20392951
ClarkBJBassettJPWangSSet al (2010) Impaired head direction cell representation in the anterodorsal thalamus after lesions of the retrosplenial cortex. Journal of Neuroscience 30(15): 5289–5302.20392951, ClarkBJBassettJPWangSSet al (2010) Impaired head direction cell representation in the anterodorsal thalamus after lesions of the retrosplenial cortex. Journal of Neuroscience 30(15): 5289–5302.20392951
( YoderRMClarkBJTaubeJS (2011) Origins of landmark encoding in the brain. Trends in Neurosciences 34(11): 561–571.21982585)
YoderRMClarkBJTaubeJS (2011) Origins of landmark encoding in the brain. Trends in Neurosciences 34(11): 561–571.21982585
YoderRMClarkBJTaubeJS (2011) Origins of landmark encoding in the brain. Trends in Neurosciences 34(11): 561–571.21982585, YoderRMClarkBJTaubeJS (2011) Origins of landmark encoding in the brain. Trends in Neurosciences 34(11): 561–571.21982585
Patricia Preston-Ferrer, Stefano Coletta, Markus Frey, Andrea Burgalossi (2016)
Anatomical organization of presubicular head-direction circuits
eLife, 5
( BrandonMPBogaardARAndrewsCMet al (2012) Head direction cells in the postsubiculum do not show replay of prior waking sequences during sleep. Hippocampus 22(3): 604–618.21509854)
BrandonMPBogaardARAndrewsCMet al (2012) Head direction cells in the postsubiculum do not show replay of prior waking sequences during sleep. Hippocampus 22(3): 604–618.21509854
BrandonMPBogaardARAndrewsCMet al (2012) Head direction cells in the postsubiculum do not show replay of prior waking sequences during sleep. Hippocampus 22(3): 604–618.21509854, BrandonMPBogaardARAndrewsCMet al (2012) Head direction cells in the postsubiculum do not show replay of prior waking sequences during sleep. Hippocampus 22(3): 604–618.21509854
( TaubeJSMullerRURanckJBJr (1990b) Head-direction cells recorded from the postsubiculum in freely moving rats II: Effects of environmental manipulations. Journal of Neuroscience 10(2): 436–447.2303852)
TaubeJSMullerRURanckJBJr (1990b) Head-direction cells recorded from the postsubiculum in freely moving rats II: Effects of environmental manipulations. Journal of Neuroscience 10(2): 436–447.2303852
TaubeJSMullerRURanckJBJr (1990b) Head-direction cells recorded from the postsubiculum in freely moving rats II: Effects of environmental manipulations. Journal of Neuroscience 10(2): 436–447.2303852, TaubeJSMullerRURanckJBJr (1990b) Head-direction cells recorded from the postsubiculum in freely moving rats II: Effects of environmental manipulations. Journal of Neuroscience 10(2): 436–447.2303852
Andrej Bicanski, N. Burgess (2016)
Environmental Anchoring of Head Direction in a Computational Model of Retrosplenial Cortex
The Journal of Neuroscience, 36
Hippocampal place cell instability after lesions of the head direction cell network
(2001)
Frontiers Integrative Neuroscience
J. Taube, R. Muller (1998)
Comparisons of head direction cell activity in the postsubiculum and anterior thalamus of freely moving rats
Hippocampus, 8
J. Goodridge, Paul Dudchenko, K. Worboys, E. Golob, J. Taube, Kimberly Wor, Jeremy Good, Ann Clark, Julia Marx, Joshua Platt (1998)
Cue control and head direction cells.
Behavioral neuroscience, 112 4
(2009)
CircStat: A MATLAB toolbox for circular statistics
( CaltonJLStackmanRWGoodridgeJPet al (2003) Hippocampal place cell instability after lesions of the head direction cell network. Journal of Neuroscience 23(30): 9719–9731.14585999)
CaltonJLStackmanRWGoodridgeJPet al (2003) Hippocampal place cell instability after lesions of the head direction cell network. Journal of Neuroscience 23(30): 9719–9731.14585999
CaltonJLStackmanRWGoodridgeJPet al (2003) Hippocampal place cell instability after lesions of the head direction cell network. Journal of Neuroscience 23(30): 9719–9731.14585999, CaltonJLStackmanRWGoodridgeJPet al (2003) Hippocampal place cell instability after lesions of the head direction cell network. Journal of Neuroscience 23(30): 9719–9731.14585999
R. Knight, C. Piette, H. Page, D. Walters, E. Marozzi, M. Nardini, S. Stringer, K. Jeffery (2014)
Weighted cue integration in the rodent head direction system
Philosophical Transactions of the Royal Society B: Biological Sciences, 369
A. Peyrache, M. Lacroix, Peter Petersen, G. Buzsáki (2015)
Internally-organized mechanisms of the head direction sense
Nature neuroscience, 18
S. Wiener, J. Taube (2005)
Head direction cells and the neural mechanisms of spatial orientation
Michaël Zugaro, A. Arleo, A. Berthoz, S. Wiener (2003)
Rapid Spatial Reorientation and Head Direction Cells
The Journal of Neuroscience, 23
( BassettJPZugaroBMuirGMet al (2005) Passive movements of the head do not abolish anticipatory firing properties of head direction cells. Journal of Neurophysiology 93(3): 1304–1316.15469962)
BassettJPZugaroBMuirGMet al (2005) Passive movements of the head do not abolish anticipatory firing properties of head direction cells. Journal of Neurophysiology 93(3): 1304–1316.15469962
BassettJPZugaroBMuirGMet al (2005) Passive movements of the head do not abolish anticipatory firing properties of head direction cells. Journal of Neurophysiology 93(3): 1304–1316.15469962, BassettJPZugaroBMuirGMet al (2005) Passive movements of the head do not abolish anticipatory firing properties of head direction cells. Journal of Neurophysiology 93(3): 1304–1316.15469962
( SongPWangX-J (2005) Angular path integration by moving “hill of activity”: A spiking neuron model without recurrent excitation of the head-direction system. Journal of Neuroscience 25(4): 1002–1014.15673682)
SongPWangX-J (2005) Angular path integration by moving “hill of activity”: A spiking neuron model without recurrent excitation of the head-direction system. Journal of Neuroscience 25(4): 1002–1014.15673682
SongPWangX-J (2005) Angular path integration by moving “hill of activity”: A spiking neuron model without recurrent excitation of the head-direction system. Journal of Neuroscience 25(4): 1002–1014.15673682, SongPWangX-J (2005) Angular path integration by moving “hill of activity”: A spiking neuron model without recurrent excitation of the head-direction system. Journal of Neuroscience 25(4): 1002–1014.15673682
( TaubeJS (2007) The head direction signal: Origins and sensory-motor integration. Annual Review of Neuroscience 30(1): 181–207.)
TaubeJS (2007) The head direction signal: Origins and sensory-motor integration. Annual Review of Neuroscience 30(1): 181–207.
TaubeJS (2007) The head direction signal: Origins and sensory-motor integration. Annual Review of Neuroscience 30(1): 181–207., TaubeJS (2007) The head direction signal: Origins and sensory-motor integration. Annual Review of Neuroscience 30(1): 181–207.
Johannes Seelig, V. Jayaraman (2015)
Neural dynamics for landmark orientation and angular path integration
Nature, 521
J. Goodridge, J. Taube (1997)
Interaction between the Postsubiculum and Anterior Thalamus in the Generation of Head Direction Cell Activity
The Journal of Neuroscience, 17
( BerensP (2009) CircStat: A MATLAB toolbox for circular statistics. Journal of Statistical Software 31(10): 1–21.)
BerensP (2009) CircStat: A MATLAB toolbox for circular statistics. Journal of Statistical Software 31(10): 1–21.
BerensP (2009) CircStat: A MATLAB toolbox for circular statistics. Journal of Statistical Software 31(10): 1–21., BerensP (2009) CircStat: A MATLAB toolbox for circular statistics. Journal of Statistical Software 31(10): 1–21.
( KnightRPietteCEPageHet al (2014) Weighted cue integration in the rodent head direction system. Philosophical Transaction of the Royal Society of London, Series B: Biological Science 369(1635): 20120512.)
KnightRPietteCEPageHet al (2014) Weighted cue integration in the rodent head direction system. Philosophical Transaction of the Royal Society of London, Series B: Biological Science 369(1635): 20120512.
KnightRPietteCEPageHet al (2014) Weighted cue integration in the rodent head direction system. Philosophical Transaction of the Royal Society of London, Series B: Biological Science 369(1635): 20120512., KnightRPietteCEPageHet al (2014) Weighted cue integration in the rodent head direction system. Philosophical Transaction of the Royal Society of London, Series B: Biological Science 369(1635): 20120512.
Pengcheng Song, Xiao-Jing Wang (2005)
Angular Path Integration by Moving “Hill of Activity”: A Spiking Neuron Model without Recurrent Excitation of the Head-Direction System
The Journal of Neuroscience, 25
( BlairHTLipscombBWSharpPE (1997) Anticipatory time intervals of head-direction cells in the anterior thalamus of the rat: Implications for path integration in the head-direction circuit. Journal of Neurophysiology 78(1): 145–159.9242269)
BlairHTLipscombBWSharpPE (1997) Anticipatory time intervals of head-direction cells in the anterior thalamus of the rat: Implications for path integration in the head-direction circuit. Journal of Neurophysiology 78(1): 145–159.9242269
BlairHTLipscombBWSharpPE (1997) Anticipatory time intervals of head-direction cells in the anterior thalamus of the rat: Implications for path integration in the head-direction circuit. Journal of Neurophysiology 78(1): 145–159.9242269, BlairHTLipscombBWSharpPE (1997) Anticipatory time intervals of head-direction cells in the anterior thalamus of the rat: Implications for path integration in the head-direction circuit. Journal of Neurophysiology 78(1): 145–159.9242269
( KadirSNGoodmanDFMHarrisKD (2014) High-dimensional cluster analysis with the masked EM algorithm. Neural Computation 26(11): 2379–2394.25149694)
KadirSNGoodmanDFMHarrisKD (2014) High-dimensional cluster analysis with the masked EM algorithm. Neural Computation 26(11): 2379–2394.25149694
KadirSNGoodmanDFMHarrisKD (2014) High-dimensional cluster analysis with the masked EM algorithm. Neural Computation 26(11): 2379–2394.25149694, KadirSNGoodmanDFMHarrisKD (2014) High-dimensional cluster analysis with the masked EM algorithm. Neural Computation 26(11): 2379–2394.25149694
The neural encoding of self-generated and externally applied movement: Implications for the perception of self-motion and spatial memory
( TaubeJSMullerRU (1998) Comparisons of head direction cell activity in the postsubiculum and anterior thalamus of freely moving rats. Hippocampus 8(2): 87–108.9572715)
TaubeJSMullerRU (1998) Comparisons of head direction cell activity in the postsubiculum and anterior thalamus of freely moving rats. Hippocampus 8(2): 87–108.9572715
TaubeJSMullerRU (1998) Comparisons of head direction cell activity in the postsubiculum and anterior thalamus of freely moving rats. Hippocampus 8(2): 87–108.9572715, TaubeJSMullerRU (1998) Comparisons of head direction cell activity in the postsubiculum and anterior thalamus of freely moving rats. Hippocampus 8(2): 87–108.9572715
M. Abeles (1982)
Quantification, smoothing, and confidence limits for single-units' histograms
Journal of Neuroscience Methods, 5
( AbelesM (1982) Quantification, smoothing, and confidence limits for single-units’ histograms. Journal of Neuroscience Methods 5(4): 317–325.6285087)
AbelesM (1982) Quantification, smoothing, and confidence limits for single-units’ histograms. Journal of Neuroscience Methods 5(4): 317–325.6285087
AbelesM (1982) Quantification, smoothing, and confidence limits for single-units’ histograms. Journal of Neuroscience Methods 5(4): 317–325.6285087, AbelesM (1982) Quantification, smoothing, and confidence limits for single-units’ histograms. Journal of Neuroscience Methods 5(4): 317–325.6285087
Transformation of head-direction signal into spatial code. bioRxiv
F. Cacucci, C. Lever, Thomas Wills, N. Burgess, J. O’Keefe (2004)
Theta-Modulated Place-by-Direction Cells in the Hippocampal Formation in the Rat
The Journal of Neuroscience, 24
A. Peyrache, Natalie Schieferstein, G. Buzsáki (2016)
Transformation of the head-direction signal into a spatial code
Nature Communications, 8
Head-direction cells in the deep layers of the dorsal presubiculum in freely moving rats
J. Taube (2009)
Head direction cells
Scholarpedia, 4
N. Kononenko, M. Witter (2012)
Presubiculum layer III conveys retrosplenial input to the medial entorhinal cortex
Hippocampus, 22
( JacobPYCasaliGSpieserLet al (2016) An independent, landmark-dominated head direction signal in dysgranular retrosplenial cortex. Nature Neuroscience 20(2): 173–175.27991898)
JacobPYCasaliGSpieserLet al (2016) An independent, landmark-dominated head direction signal in dysgranular retrosplenial cortex. Nature Neuroscience 20(2): 173–175.27991898
JacobPYCasaliGSpieserLet al (2016) An independent, landmark-dominated head direction signal in dysgranular retrosplenial cortex. Nature Neuroscience 20(2): 173–175.27991898, JacobPYCasaliGSpieserLet al (2016) An independent, landmark-dominated head direction signal in dysgranular retrosplenial cortex. Nature Neuroscience 20(2): 173–175.27991898
J. Taube, Heather Burton (1995)
Head direction cell activity monitored in a novel environment and during a cue conflict situation.
Journal of neurophysiology, 74 5
( GolobEJStackmanRWWongACet al (2001) On the behavioral significance of head direction cells: Neural and behavioral dynamics during spatial memory tasks. Behavioral Neuroscience 115(2): 285–304.11345955)
GolobEJStackmanRWWongACet al (2001) On the behavioral significance of head direction cells: Neural and behavioral dynamics during spatial memory tasks. Behavioral Neuroscience 115(2): 285–304.11345955
GolobEJStackmanRWWongACet al (2001) On the behavioral significance of head direction cells: Neural and behavioral dynamics during spatial memory tasks. Behavioral Neuroscience 115(2): 285–304.11345955, GolobEJStackmanRWWongACet al (2001) On the behavioral significance of head direction cells: Neural and behavioral dynamics during spatial memory tasks. Behavioral Neuroscience 115(2): 285–304.11345955
( PeyracheARouxLSchiefersteinNet al (2016) Transformation of head-direction signal into spatial code. bioRxiv. Epub ahead of print 20 9 DOI: 10.1101/075986.)
PeyracheARouxLSchiefersteinNet al (2016) Transformation of head-direction signal into spatial code. bioRxiv. Epub ahead of print 20 9 DOI: 10.1101/075986.
PeyracheARouxLSchiefersteinNet al (2016) Transformation of head-direction signal into spatial code. bioRxiv. Epub ahead of print 20 9 DOI: 10.1101/075986., PeyracheARouxLSchiefersteinNet al (2016) Transformation of head-direction signal into spatial code. bioRxiv. Epub ahead of print 20 9 DOI: 10.1101/075986.
J. Taube (2007)
The head direction signal: origins and sensory-motor integration.
Annual review of neuroscience, 30
H. Blair, P. Sharp (1995)
Anticipatory head direction signals in anterior thalamus: evidence for a thalamocortical circuit that integrates angular head motion to compute head direction
, 15
( PeyracheALacroixMMPetersenPCet al (2015) Internally organized mechanisms of the head direction sense. Nature Neuroscience 18(4): 569–575.25730672)
PeyracheALacroixMMPetersenPCet al (2015) Internally organized mechanisms of the head direction sense. Nature Neuroscience 18(4): 569–575.25730672
PeyracheALacroixMMPetersenPCet al (2015) Internally organized mechanisms of the head direction sense. Nature Neuroscience 18(4): 569–575.25730672, PeyracheALacroixMMPetersenPCet al (2015) Internally organized mechanisms of the head direction sense. Nature Neuroscience 18(4): 569–575.25730672
W. Skaggs, B. McNaughton, K. Gothard (1992)
An Information-Theoretic Approach to Deciphering the Hippocampal Code
B. Clark, J. Bassett, SarahS Wang, J. Taube (2010)
Impaired Head Direction Cell Representation in the Anterodorsal Thalamus after Lesions of the Retrosplenial Cortex
The Journal of Neuroscience, 30
( CaltonJLTurnerCSCyrenneD-LMLeeBRTaubeJS (2008) Landmark control and updating of self-movement cues are largely maintained in head direction cells after lesions of the posterior parietal cortex. Behavioral Neuroscience 122(4): 827–840.18729636)
CaltonJLTurnerCSCyrenneD-LMLeeBRTaubeJS (2008) Landmark control and updating of self-movement cues are largely maintained in head direction cells after lesions of the posterior parietal cortex. Behavioral Neuroscience 122(4): 827–840.18729636
CaltonJLTurnerCSCyrenneD-LMLeeBRTaubeJS (2008) Landmark control and updating of self-movement cues are largely maintained in head direction cells after lesions of the posterior parietal cortex. Behavioral Neuroscience 122(4): 827–840.18729636, CaltonJLTurnerCSCyrenneD-LMLeeBRTaubeJS (2008) Landmark control and updating of self-movement cues are largely maintained in head direction cells after lesions of the posterior parietal cortex. Behavioral Neuroscience 122(4): 827–840.18729636

Publisher: SAGE
Copyright: Copyright © 2022 by SAGE Publications Ltd and British Neuroscience Association, unless otherwise noted. Manuscript content on this site is licensed under Creative Commons Licenses
ISSN: 2398-2128
eISSN: 2398-2128
DOI: 10.1177/2398212817721859
Publisher site: See Article on Publisher Site

Abstract

Background: Visual landmarks are used by head direction (HD) cells to establish and help update the animal’s representation of head direction, for use in orientation and navigation. Two cortical regions that are connected to primary visual areas, postsubiculum (PoS) and retrosplenial cortex (RSC), possess HD cells: we investigated whether they differ in how they process visual landmarks. Methods: We compared PoS and RSC HD cell activity from tetrode-implanted rats exploring an arena in which correct HD orientation required discrimination of two opposing landmarks having high, moderate or low discriminability. Results: RSC HD cells had higher firing rates than PoS HD cells and slightly lower modulation by angular head velocity, and anticipated actual head direction by ~48 ms, indicating that RSC spiking leads PoS spiking. Otherwise, we saw no differences in landmark processing, in that HD cells in both regions showed equal responsiveness to and discrimination of the cues, with cells in both regions having unipolar directional tuning curves and showing better discrimination of the highly discriminable cues. There was a small spatial component to the signal in some cells, consistent with their role in interacting with the place cell navigation system, and there was also slight modulation by running speed. Neither region showed theta modulation of HD cell spiking. Conclusions: That the cells can immediately respond to subtle differences in spatial landmarks is consistent with rapid processing of visual snapshots or scenes; similarities in PoS and RSC responding may be due either to similar computations being performed on the visual inputs, or to rapid sharing of information between these regions. More generally, this two-cue HD cell paradigm may be a useful method for testing rapid spontaneous visual discrimination capabilities in other experimental settings. Keywords Head direction cells, postsubiculum, retrosplenial cortex, landmarks, visual discrimination, spatial memory, in vivo rodent electrophysiology Received: 17 April 2017; accepted: 28 June 2017 Introduction How the brain forms a representation of external, navigable space same relative PFDs, even if the orientation of the entire cell popula- is a current area of intense enquiry because it involves transforma- tion changes from one environment to the next. Local environmen- tion of sensory inputs into higher-order, more abstract cognitive tal landmarks – mainly visual – establish the population orientation structures, and thus has wide relevance to cognition generally. when the animal enters an environment, and the signal is updated as One of the foundations of the place representation is the ‘sense of the animal moves around by means of self-motion information such direction’, supported by a network of brain regions known as the as vestibular, optic flow, motor efference and proprioceptive cues to head direction (HD) system, which uses previously learned visual movement (Taube, 2007; Yoder et al., 2011). Although HD cells landmarks to re-orient when an animal re-enters a familiar envi- typically begin firing essentially immediately on entry into a famil- ronment. This study investigates the neural basis of this rapid ori- iar environment (Jankowski et al., 2014), the environmental cues entation process, which is important for understanding how perception and memory processes shape the place representation. Division of Psychology and Language Sciences, University College The HD system in rodents (and probably all vertebrates) con- London, London, UK tains so-called HD cells, which fire when the animal faces in a par- ticular direction, regardless of position, and are close to silent Corresponding author: otherwise (Taube et al., 1990a; 1990b). Each cell has its own pre- Kate Jeffery, Division of Psychology and Language Sciences, University ferred firing direction (PFD) in a given environment, and the College London, 26 Bedford Way, London WC1H 0AP, UK. ensemble of HD cells is coherent – that is, the cells maintain the Email: k.jeffery@ucl.ac.uk Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License (http://www.creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage). 2 Brain and Neuroscience Advances are learned, not hard-wired (Taube and Burton, 1995), which means PoS rat had received a saline sham injection into the lateral genicu- that some kind of rapid recognition process must take place. How late nucleus (LGN), as part of a different study. The rats were housed and where this occurs is not known, but is likely to be in HD regions in individual cages in a temperature and humidity-controlled colony close to the visual system. room that was kept on an 11:11 h light:dark cycle plus 1 h each of Current evidence suggests that the postsubiculum (PoS) and the half-light simulated dawn and dusk. All the rats had free access to retrosplenial cortex (RSC) have an important role in processing food and water prior to surgery and began food restriction to main- landmark information derived from visual areas and relaying this tain 90% of their free-feeding weight 1 week after surgery. All proce- input to interconnected areas of the HD cell circuit and the hip- dures were licensed by the UK Home Office following the revised pocampal formation (for review, see Yoder et al., 2011). Consistent ASPA regulations (2013) modified by the European Directive with this hypothesis, Goodridge and Taube (1997) found that tuning 2010/63/EU. Two of the RSC-implanted rats also took part in curve precision (width of the tuning curve) and stability of anterior another experiment, in a different apparatus (Jacob et al., 2016). thalamic HD signals were impaired after PoS lesions, and the cells showed reduced responsiveness to rotations of the landmarks. Yoder et al. (2015) reported a similar effect on lateral mammillary HD neu- Electrodes and surgery rons, as did Calton et al. (2003) on CA1 place fields, downstream of Each rat was anesthetised and implanted with a microdrive the HD signal. Similarly, Clark et al. (2010) found the same effects (Axona) that was configured with either four or eight tetrodes on both tuning curve width/stability and landmark control following threaded inside a guide cannula. Each tetrode was made of indi- RSC lesions, although to a lesser extent. In contrast, lesions of the vidual 90:10 platinum–iridium wires (California Fine Wire) of parietal (Calton et al., 2008) or the postrhinal (Peck and Taube, diameter 25 µm (for the four-tetrode drives) or 17 µm diameter 2017) cortex did not impair landmark control of anterior dorsal thal- (for the eight-tetrode drives). The tips of the electrodes were amus (ADN) HD cells, suggesting that the processing of visual land- plated in a 1:9 0.5% gelatine:Kohlrausch platinum solution to mark information is routed via the PoS and the RSC. These studies approximately 250 kΩ using a pulse generator (Thurlby Thandar collectively point to a likely role for PoS and RSC in passing infor- TGP-110) and a current source amplifier (A.M.P.I. ISO-FLEX) mation about landmarks (or at least their visual properties) to the that delivered 2 µA current for 550 ms to each channel. The subcortical circuits that collate the HD information and generate a microdrive and the guide cannula were fastened to the skull with stable signal. It is thus of interest to look for possible differences in dental acrylic (Simplex Rapid) covering seven supporting the contributions made by the two regions. screws of 1.6 × 3 mm (Small Parts) that were inserted into the It is not known why there should be two separate landmark pro- occipital, parietal and frontal cranial bones. One of the support- cessing regions, but one possibility is that they differ in some aspect ing screws made contact with the frontal cortex and was con- of their contribution to the processing of environmental landmarks. nected to the microdrive ground wire. We therefore set out to compare the responses of these neurons dur- Implant coordinates were based on previous studies of HD ing re-orientation in a situation where detailed landmark-cue pro- cells in the RSC (Cho and Sharp, 2001) and the PoS (Taube cessing is required. We recorded HD neurons as rats foraged in a et al., 1990a); for RSC (n = 12 rats; n = 6 left hemisphere, n = 6 cylinder within a curtained enclosure (Figure 1(a)). The infinite rota- right hemisphere), these were 5.4 mm posterior to bregma, 0.6– tional symmetry of this arrangement (Figure 1(b)) was broken by a 0.8 mm lateral to the midline and 0.2–0.9 mm ventral from the pair of cue cards on the cylinder wall, located opposite each other; cortical surface, while two sets of coordinates were used for the the resulting twofold symmetry of the cue pair (Figure 1(c)) was PoS (n = 5 rats; all in the left hemisphere); being, respectively, further broken by making the cards different so that the environment 6.7 or 7.5 mm posterior to bregma, 2.8 or 3.2 mm lateral to the was now polarised, provided the cues could be discriminated (Figure midline and 1.6 or 1.9 mm ventral from the cortical surface. 1(d)). The discriminability of the cue card pair varied between high, After surgery, the animals were monitored until they awoke, and moderate and low (Figure 1(e)) – in the HIGH condition, one card meloxicam (Metacam) was given in jelly for three consecutive was black and one white; in the MOD condition, a black bar was days as pain relief. All animals were allowed to recover for oriented and/or positioned one way on one card and the other way on 1 week prior to the start of recording. the other, and in the LOW condition, the cards were visually identi- cal (although they may have been distinguishable by smell; hence discriminability is assumed to be low and not zero). Cue control apparatus PoS and RSC HD cells were recorded in sessions ranging from 4 to 12 trials; between trials, the rat was removed and mildly The cue control experiments were performed inside a cylindri- disoriented, and the entire environment was frequently also cal arena (diameter, 74 cm; height, 50 cm) made of plywood rotated to disconnect it from static external cues. Cells were painted with light grey matt acrylic and placed at the centre of tested for basic firing parameters (firing rate, spiking characteris- a black curtained enclosure (diameter, 260 cm) (Figure 1(a)). A tics, and possible spatial localisation) and to determine whether cylinder was selected as a recording arena to minimise the their firing directions rotated appropriately with the cue pair, influence of the environment’s geometry as an orienting cue indicating successful detection and discrimination. (Golob et al., 2001; Knight et al., 2011). Attached to the inner wall of the cylinder with Velcro tape were two 50 × 50 cm cue cards (Figure 1(b)) made from black and/or white polypropyl- Materials and methods ene sheets, each subtending ~77° of arc, and located 180° apart. Two of the cue pairs were plain – either both black (iden- Subjects tical-cue controls; B-B) or one black and one white (a maxi- A total of 18 adult male Lister Hooded rats weighing between 317 mally salient high-contrast condition; B-W). The remaining and 437 g at the time of surgery were used for the experiments. One cue pairs, with one exception, were made from white card Lozano et al. 3 Figure 1. Cue-discrimination setup for recording HD cells in the RSC and PoS of freely moving rats. (a) Recording environment. Rats foraged for rice in a 50 cm high by 74 cm diameter cylindrical arena with two cue cards 50 × 50 cm attached to the inner wall 180° apart. The arena was situated in the centre of a 260 cm diameter, black circular curtained enclosure. (b)–(d) Environmental symmetry, and hence HD cell anchoring, varies as a function of cue perception. (b) If no cues are detected then the environment has infinite rotational symmetry, and HD orientations might be random. (c) If the cue pair is detected by the cells but not discriminated, then the environment has twofold rotational symmetry and an HD cell might fire in either of two directions, at random. (d) If the cues are both detected and discriminated, then the environment lacks rotational symmetry (is polarised) and an HD cell would be able to fire in a constant direction. (e) Cue card patterns, grouped into discriminability categories (high, moderate and low), and their corresponding abbreviations. B-W: black–white, V-H: vertical–horizontal, L-R: left–right, T-B: top–bottom, B-B: black–black and L-L are both L-shaped. The V-H cards were also sometimes used in white-on-black configuration. decorated with a black bar of 14 × 50 cm; these were thus equal the exceptional pair, black and white were reversed for the ori- in overall luminance and contrast, so discrimination would entation cues. require processing of the cues’ internal structure. The position The arena rested over a black vinyl sheet and was lit from of the bar in each pair differed in, respectively, orientation above by six light fixtures that provided approximately 250 lux (vertical vs horizontal; V-H cues), lateral position (left vs right; of light, with a radio attached to the ceiling as a source of white L-R cues) or vertical position (top vs bottom; T-B cues). For noise to reduce the effect of directional auditory cues. 4 Brain and Neuroscience Advances encourage the rat to sample all the locations and facing directions Screening/recording procedures within the cylinder. With the exception of the black–white (B-W) Signal processing and tracking. Single units were recorded cue pair, which was used to establish cue control, the order of using a headstage amplifier connected to a microdrive and con- presentation for the different cue card stimuli was pseudo-ran- nected to the multichannel recording system (DacqUSB, Axona) domised between rats to prevent temporal effects due to the via a flexible lightweight tether. Traces from individual channels changing experience of the animal across days. were collected at a sampling rate of 48 kHz, amplified 6000– For one rat, a control procedure was conducted to assess the 20,000 times and band-pass filtered from 300 Hz to 7 kHz. The relative contribution of vision and olfaction to the landmark dis- traces from each electrode were referenced against an electrode crimination. Using the patterned cues that had rotational symme- from a different tetrode that showed low spiking activity. Spikes try (V-H, T-B and L-R), the visual identity of the cards was were defined as short-lasting events that crossed a user-defined reversed by rotating around their central points, 90° for V-H and threshold; the period from 200 µs before to 800 µs after threshold- 180° for T-B and L-R, to determine whether firing remained crossing were captured and saved along with the corresponding aligned relative to the physical cards or to the visual appearance. timestamps. To track the rat’s head position and facing direction in the horizontal plane, two arrays of light-emitting diodes (LEDs), Data analysis one large and one small, were transversely positioned 8 cm apart on a stalk connected to the headstage. The LED positions were Spike-sorting was performed using KlustaKwik (Kadir et al., recorded at a sampling rate of 50 Hz by a camera attached to the 2014) followed by manual refinement using the TINT software ceiling. Spike times, the rat’s head position in x- and y-coordi- package (Axona). Polar plots were generated, and cells that nates and its heading in degrees were saved for offline analysis. showed directional firing by eye were selected for further quanti- tative analysis, which included extracting peak firing rate and directional clustering – Rayleigh vector (R-vector) – scores. HD cell screening. The screening sessions for single units tuned Cells that were recorded on the same tetrode across sessions were to HD were conducted inside a 76 × 76 × 50 cm box in a room treated as unique if more than 5 days separated the sessions. For separate from where the cue control experiments took place. The each trial, a cell was accepted into further analysis if it had a peak animals had visual access to a polarising cue card placed inside firing rate >1.0 Hz across sessions, a peak:mean rate ratio > 2 and the box, and distal room cues. The screening sessions consisted a R-vector p-value <0.05. Cells were excluded altogether if only of 5–10 min trials during which the rat foraged for rice while three trials or fewer per session remained after selection. spikes were monitored. Polar plots were generated with spike- Data were grouped in three ways: (a) Those pertaining to sorted data using the software TINT (Axona) and examined to cell-specific characteristics, such as firing rate, R-vector score determine whether single units were tuned to HD. In screening and tuning curve width were averaged across a given session. sessions during which HD cells were not found, the tetrodes were (b) Data concerning trial-specific characteristics such as firing lowered by ~50 µm and the rat was screened again 4 h later or the direction relative to the cue card array were averaged across a given next day. cell. (c) Data concerning session-specific characteristics, such as the spread of firing directions, were averaged across a session. Cue control procedure. After isolating single units tuned to HD The basic protocol for firing direction analysis is shown in Figure during screening, the rat was taken to the recording room inside a 2. First, a tuning curve was derived (see below). Each tuning curve closed opaque box. A cue control session was conducted using a was recorded in the camera frame of reference, within which the similar recording procedure as described previously (Knight cues rotated, but it was necessary to align these within a common et al., 2011). Briefly, each session started with a series of baseline reference frame so that population statistics could be derived. To do trials (2–4 trials) where the two cue cards remained aligned in the this, the tuning curves were first specified relative to one of the two same location relative to the room coordinates. This was followed cues (‘Cue 1’; Figure 2(b)) and the mean firing direction across the by a series of rotation trials (4–8 trials) during which the cue cards session computed. Then, the tuning curves were realigned relative to were rotated together by ±45°, ±90°, ±135° or 180° (Figure 1(c)) this mean; this value (deviation from the session mean) became the to test whether HD cells could discriminate and use the visual value that entered into the population analysis. Finally, the angles features of the cue cards as orienting landmarks (Figure 1(d)). For were doubled in order to remove any bipolarity that might be present each recording session, the starting location of the cue cards as due to cue confusion – this was done so as to enable determination well as the magnitude and direction of the cue rotations were of cue use independently of cue discrimination. pseudo-randomised. The length of each trial was 300 s and was initiated via remote control after placing the rat inside the cylinder with a pseudo-random location and facing direction. Prior to each Cell-specific firing characteristics trial, the cue cards, recording arena and the base of the cylinder were inverted and wiped with 75% ethanol to scramble olfactory Data were analysed using MATLAB (MathWorks) with cus- cues. During the inter-trial interval, the rats remained inside a tom-made programs and functions taken from the CircStat tool- holding box outside the curtained enclosure, and then prior to box (Berens, 2009). To analyse the firing characteristics of being replaced in the recording box they were mildly disoriented single units, the spike times and position samples of the rat’s by being passively transported and rotated in the holding box facing direction were sorted into bins of 6°. The mean firing around the periphery of the curtained enclosure, thereby prevent- rate per HD bin (Hz) was calculated by taking the sum of the ing the animals from using self-motion cues to track their orienta- spikes divided by the total amount of time (s) that the rat’s fac- tion. During the recording trials, the experimenter remained ing direction was located in each bin. A 5-bin (30°) smoothing outside the curtained enclosure, tossing rice into the arena to kernel was applied to the circular histogram of the firing rate as Lozano et al. 5 Figure 2. Tuning curve analysis. (a) Example of the tuning curve derived from a single PoS HD neuron in a series of recording trials, selected from a 12-trial session. The plots show firing rate as a function of HD, normalised to the peak rate (shown). (b) Schematic from a hypothetical set of trials showing how data were transformed into a common reference frame to allow population comparisons. The original reference frame was that of the room-fixed camera: the top row shows an idealised trial in which the cell consistently rotated its firing with the cue pair, but was aligned relative to Cue 1 on trials 1, 2 and 5 and 180° away from it on trials 3 and 4. The bottom row shows these data expressed in the different reference frames: in the camera frame, the firing direction is scattered; relative to Cue 1, it is bipolar with three trials in one direction and two in the opposing direction; relative to the session mean (arrow), the bulk of the firing is now aligned to zero, and after angle-doubling the bipolar distribution is now transformed to unipolar. a function of HD to minimise random influences to the firing the effect of wrapping the 180° points around to zero, thus render- rate in each cell (Abeles, 1982). Polar plots of the smoothed ing a bipolar distribution unipolar for the purposes of analysis. histograms were generated to visualise the cell’s tuning curve (Figure 2(a)). Test for bipolar tuning curves. In order to test whether there The parameters that were used to quantify the tuning curve was any within-trial tendency towards bipolarity, due to (per- characteristics for each cell were firing rate, PFD, tuning width, haps) intermittent reversing of the tuning curve arising from con- and mean R-vector length. The firing rate of the HD cell was fusion between the cues, a test for bipolarity was conducted by defined as the bin with the maximum firing rate; that is, the mode running a circular autocorrelation on the smoothed binned data of the circular histogram, and the PFD defined to be the direc- from each trial, using the MATLAB circshift and corr functions. tional bin of the mode. The tuning width, or directional firing From the autocorrelation, the values at 90° and 270° relative to range of the cell, was determined by computing two standard the peak were extracted and averaged, and compared with the deviations from the circular mean direction. The R-vector score peak at 180° using a one-tailed t-test of the hypothesis that the was used as a measure of the directional tuning in each cell. 180° peak would be larger. Values of R close to 1 indicate that the spikes are closely clus- tered around a single value and values close to 0 indicate that the spikes were distributed around all facing directions. Spatial firing patterns. In order to determine whether there To examine how a cell behaved in relation to the cue card pair was any spatial specificity to the firing, which could be of com- across trials, the PFDs were realigned according to their relation- putational utility (Bicanski and Burgess, 2016), and has been ship to one of the two cues (arbitrarily called ‘Cue 1’) and the reported for some types of HD cells in cortical regions (Cacucci circular mean of the session values was computed. Each PFD et al., 2004; Peyrache et al., 2016), we examined the spatial dis- was then expressed as a deviation from this mean. tribution of spikes in the same way as is usually done for place In some cases, it was necessary to test whether a low R-vector cells, extracting the spatial information content of the firing, the score could be due to bipolarity of the firing distribution. To do coverage relative to the whole environment and the coherence of this, the values were doubled and plotted modulo 360 – this has the firing distribution. Path data were extracted by smoothing the 6 Brain and Neuroscience Advances position points with a 400 ms window and first plotting the spikes compensate for variability in firing rate between cells. Running at their corresponding locations, for visual inspection. These data speeds below 2 cm/s were excluded from running speed analy- were then used to generate dwell-time-normalised firing rate sis. Linear running speed was binned in intervals of 2 cm/s, and maps by binning the spike and position data into 3 cm bins and AHV was binned in intervals of 2°/s. The firing rate was calcu- dividing the firing rate by dwell time, and then smoothing the lated by counting the spikes in each bin and dividing by the time map with a boxcar of three spatial bins (whereby the value in spent in that bin (dwell time) and then normalised to the peak for each bin was replaced by the average for that bin plus the sur- that trial to enable comparison across trials/cells. Bins with rounding eight bins). To eliminate sampling bias due to the rat’s dwell times of less than 1.5% total trial time or with fewer than inability to face all directions everywhere in the arena, we under- five spikes were discarded; a linear regression was run on the took the spatial analysis using only the inner 50% of the arena, in remainder to generate a slope value. Because dwell time which directional sampling was homogeneous. In case some cells decreased with increasing velocity, which might cause artefacts might have place fields near the edge of the arena, we also anal- in the rate/speed relationship, the baseline for each trial was cal- ysed the hemi-cylinder lying in the direction of the cell’s PFD, culated by generating an artificial continuous 10 Hz spike train, for which the directional bias was much reduced (since it was analysing it in the same way and subtracting this control slope easy for the rat to face in the cell’s PFD in this region); this made from the raw data slope. For AHV, left and right turns were ana- little difference, so we chose the inner region as being the most lysed separately and the absolute slope values then combined for conservative measure. that cell. This is because previous recordings from other brain Spatial information in bits/spike was computed following the regions have found cells with asymmetric AHV rate profiles method of Skaggs et al. (1993). Coverage was calculated as the (Bassett and Taube, 2001), being negative in one direction and percentage bins above 20% peak rate, and coherence of spatial positive in the other, which would cancel if the raw values were firing was determined using a Pearson’s correlation between the taken. The resulting difference values were entered into a t-test smoothed and unsmoothed rate maps. Actual data were com- comparing PoS and RSC. pared against a control dataset generated by taking the spike data Anticipatory time intervals (ATIs) were estimated using a time and shifting it forwards by 1000 position points (20 s), which slide analysis in the manner of Blair and Sharp (1995). Given the would scramble any spatial specificity of firing while leaving the camera sample rate of 50 Hz, spike times were shifted forwards in temporal dynamics unchanged. Each firing rate map was com- 20 ms intervals from 20 to 160 ms. For each head turn, a tuning pared with its shuffled control map using a paired t-test. curve was constructed for leftwards and rightwards head turns in the manner described. A population vector method was used to calculate the PFD of each tuning curve (Song and Wang, 2005) as Temporal firing patterns. We looked at temporal patterns of firing including inter-spike interval (ISI) time to peak and decay time to half-peak, as well as theta-frequency rhythmicity. For the   ISI analysis, only trials with >145 spikes were used and only one rx sin( ) ii   PFD = arctan trial (usually the first) was used from each cell. A histogram of  N   rx cos( )  ISIs with 2 ms bins was generated for each cell and the peak was ii  i  taken as the centre of the bin with the highest count. We calcu- lated decay time by fitting, to the histogram, a one-term exponen- bx where r is the firing rate in bin i of N with mean bin direction x , and i i tial decay function of the form ya = from the peak to peak + 1 s, arctan denotes quadrant-specific arctangent function. using the fit function from MATLAB’s Curve Fitting toolbox. The difference between PFD for leftwards and rightwards Time to half-peak was then taken as the time taken for the expo- tuning curves was plotted, and a line was fitted to it using the nential fit to decay to half the peak value. MATLAB function polyfit. To allow for estimations of ATIs at a We measured theta modulation by plotting autocorrelograms finer scale than measurement, the ATI was taken as the value of of the spike trains over the range ±500 ms, in bins of 10 ms dura- this fitted line when PFD difference was equal to zero, extracted tion. The plots were then highly smoothed (20 bins) to remove using the MATLAB function polyval. local variations, and the values at the 7th bin from the central peak (expected trough at 60–70 ms) and the 12th bin (expected peak at 120–130 ms) determined: the theta modulation index was Stabilisation analysis taken as the difference between these values divided by their To investigate the time course of cue control establishment, we sum. If there is significant theta modulation, then the 12th bin looked at firing within the PFD range (PFD ±20°) across the trial, should be a peak and the 7th bin a trough, yielding a positive taking both the percentage of total spikes emitted in each time modulation index varying from 0 to 1. Conversely, values below decile (30 s – trials were 300 s long), and the time taken to reach zero would indicate a descending likelihood of a cell spiking with each spike count decile. time between the first and second time-points. Movement correlates. The relationship between linear or Histology angular speed and firing rate was examined by analysing those portions of the trial when the animal’s HD was within 45° either After completion of the electrophysiological recordings, the rats side of the PFD of the cell, and correlating the firing rate with were anaesthetised and killed with an overdose of sodium pento- movement speed. Correlations of firing rate with linear running barbital Euthatal, 150 mg/kg and perfused transcardially with speed and angular head velocity (AHV) were computed as per- saline followed by 4% formalin solution. A day before section- centage firing rate change as a function of movement speed, to ing, the brains were placed in 4% formalin/20% sucrose solution Lozano et al. 7 Table 1. Basic firing statistics compared between PoS and RSC HD cells. PoS (n = 74) RSC (n = 75) F/T statistic (dof) p value −7 Peak rate (Hz) 8.14 ± 0.63 22.40 ± 2.58 t(147) = 5.34 3.48 × 10 −8 Mean rate (Hz) 1.83 ± 0.19 5.07 ± 0.52 t(147) = 5.86 2.92 × 10 −5 Tuning curve width (°) 43.65 ± 1.13 49.29 ± 1.28 t(147) = 3.31 1.18 × 10 Tuning curve R-vector 0.59 ± 0.02 0.53 ± 0.02 t(147) = 1.53 0.13 −7 Spatial information (data: control 1.63 ± 0.06 2.22 ± 0.10 t(145) = 5.00 7.00 × 10 bits/spike ratio) −4 Coverage (data: control % ratio) 0.98 ± 0.01 1.48 ± 0.04 t(145) = 3.67 1.70 × 10 −4 ISI rise time to peak (ms) 13.5 ± 1.6 7.5 ± 0.4 t(144) = 3.74 2.66 × 10 −4 ISI decay time to half-peak (ms) 80.1 ± 6.5 48.8 ± 5.8 t(144) = 3.61 4.20 × 10 Theta modulation index –0.05 ± 0.03 –0.03 ± 0.00 t(148) = 2.64 0.009 Abs. slope of firing rate correlation 0.69 ± 0.1 vs 0.06 ± 0.001 0.6 ± 0.1 vs 0.07 ± 0.01 Data vs 10 Hz control < 0.0001 with linear running speed (%/m/s) F(1, 1) = 40.10 vs 10 Hz control PoS vs RSC 0.50 F(1, 1) = 0.50 Interaction 0.56 F(1, 64) = 0.30 Abs. slope of firing rate correlation 0.12 ± 0.01 vs 0.03 ± 0.05 0.08 ± 0.01 vs 0.03 ± 0.03 Data vs 10 Hz control 0.003 with angular head velocity (%/°/s) F(1, 1) = 9.70 vs 10 Hz control PoS vs RSC 0.003 F(1, 1) = 5.0 Interaction 0.025 F(1, 64) = 4.20 Anticipatory time interval (ms) 14.37 ± 8.19 47.91 ± 4.28 t(147) = 3.62 0.0002 PoS: postsubiculum; RSC: retrosplenial cortex; dof: degrees of freedom: ISI: inter-spike interval. N’s differ slightly because some of the data trials could not be curve-fitted. for cryoprotection. Coronal or sagittal sections of the brains were granular RSC as illustrated by the final tetrode depth, which cut in 40 µm sections using a microtome at −20° C and mounted could have under-estimated the actual number of HD cells. on microscopic slides (Superfrost BDH). The brain sections were Results and statistics are detailed below and summarised in Nissl stained with 0.1% cresyl violet (Sigma-Aldrich) or 0.5% Table 1. thionin (Sigma-Aldrich) and cover-slipped with DPX (Sigma- Aldrich). The slides were examined under a light microscope Cell-specific firing parameters (Leica) and imaged with a digital camera mounted on the micro- scope. Images with visible electrode tracks were saved for histo- As described in section ‘Methods’, the cell-specific character- logical analysis. The electrode location was verified by examining istics of peak and mean firing rate, R-vector score and tuning the brain region where the final electrode track was located and curve width were calculated for each trial and then averaged then relating the location to a standard rat atlas of the regions across a given session for each cell. Statistical parameters are (Paxinos and Watson, 2007). summarised in Table 1 and graphically shown in Supplementary Figure 2(A). Peak firing rates were significantly higher in RSC, being 8.14 ± 0.63 Hz in PoS and 22.40 ± 2.58 Hz in RSC Results −7 (t(147) = 5.34, p = 3.48 × 10 ). Similarly, mean firing rates were higher in RSC: 1.83 ± 0.19 in PoS and 5.07 ± 0.52 in RSC Representative example sections showing the final tetrode −8 (t(147) = 5.86, p = 2.92 × 10 ). Tuning curves were narrower placement of rats implanted in the PoS and the RSC are shown in PoS, being 43.65° ± 1.13° in PoS and 49.29° ± 1.28° in RSC in Supplementary Figure 1. A total of 149 unique cells meeting −5 (t(147) 3.31, p = 1.18 × 10 ). However, directional tuning was the criteria for HD cells (see section ‘Methods’) were recorded similar in both structures: there was no difference in R-vector from 18 rats in 78 sessions, of which 74 cells were recorded score, being 0.59 ± 0.02 in PoS and 0.53 ± 0.02 in RSC from 6 rats implanted in the PoS over 34 sessions, and 75 cells (t(147) = 1.53, p = 0.13). from 12 rats implanted in the RSC over 44 sessions. We observed that a greater proportion of RSC HD cells (73%) were found below 1500 µm which is likely in the deeper granu- Spatial firing lar region. However, although the implant target coordinates were the same for all rats, there was a degree of variability in Visual inspection of the firing rate maps revealed mostly uni- the actual anterior-posterior and medio-lateral coordinates in form firing, although there were sometimes patches of inhomo- which the tetrodes were implanted (Supplementary Figure 1). geneous firing at a single-trial level and occasional clear Furthermore, the tetrodes did not cover all the layers of the place fields of the kind seen in hippocampal recordings 8 Brain and Neuroscience Advances (Supplementary Figure 3). We quantified spatial modulation by Consistent with our observations of higher firing rates in RSC, the considering spatial information content, coverage and coher- time to the ISI peak was considerably longer in PoS (13.5 ± 1.6 ms) ence. As detailed in section ‘Methods’, only the central 50% (by than in RSC (7.5 ± 0.4 ms), these being significantly different −4 radius) of the arena was used in order to remove bias induced by (t(144) = 3.74, p = 2.66 × 10 ). The time to return to the half-peak, inhomogeneous directional sampling around the arena edges. which can be thought of as a measure of the prevalence of longer Both PoS and RSC HD neurons showed evidence of a spa- ISIs, was also different, being 80.1 ± 6.5 ms for PoS and 48.8 ± 5.8 ms −4 tial component to their firing, revealed by higher spatial infor- for RSC (t(144) = 3.61, p = 4.2 × 10 ). mation content, lower coverage and by higher coherence For each cell, autocorrelograms were generated for each trial relative to the time-shifted control data (Supplementary Figure in a session (Supplementary Figure 5(A)) and from these, a theta 2(B)). Spatial information content (bits per spike) for the PoS modulation index was calculated (see section ‘Methods’). Visual data was 0.23 ± 0.02, while for the control data it was inspection of the autocorrelograms revealed no evidence of theta 0.16 ± 0.01, which was significantly different, as revealed by a modulation, although there were frequently peaks and troughs −18 paired one-tailed t-test (t(73) = 11.57, p = 9.72 × 10 ). suggestive of slightly longer scale periodicity (see examples Corresponding values for the RSC data were 0.15 ± 0.01 and from the cell shown in the Supplementary Figure 5(A)). However, for the control data were 0.07 ± 0.00, which was significantly these patterns did not persist across trials and are likely due to the −17 different (t(73) = 11.16, p = 1.13 × 10 ). Coverage was slightly dynamics of the animal’s movements. For example, if a rat is less for both cell types in the real data; PoS coverage in the real foraging by sweeping its head back and forth, then a cell’s tuning data condition was 52 ± 3%, and in the time-shifted control curve will be visited and re-visited over a regular time period of condition was 53 ± 3% which was significantly greater (paired up to several seconds. It may be that rats varied their movement −5 one-tailed t(72) = 4.14, p = 4.70 × 10 ). For RSC, coverage for patterns across trials. the data was 62 ± 2% and for the control data was 64 ± 2%, The quantitative theta modulation index confirmed the visual −4 which was also significantly higher (t(73) = 3.58, p = 3.1 × 10 ). impression: in general, values were, if anything, negative (a Coherence values were greater for the data compared with the lower value at the expected peak than at the expected trough) and time-shifted control data. For the PoS data were 0.24 ± 0.01 slightly more so, and more dispersed, for PoS (Supplementary and for the time-shifted control data were 0.20 ± 0.01, which Figure 5(B)). The values for PoS were −0.049 ± 0.006 and for −9 was significantly lower (t(70) = 6.38, p = 8.4 × 10 ), while for RSC were −0.032 ± 0.002 (t(148) = 2.63, p = 0.005). RSC the real data had a coherence value of 0.27 ± 0.01 and the control data had a coherence of 0.19 ± 0.00, which was again −19 Movement correlates significantly lower (t(73) = 12.01, p = 2.89 × 10 ). We then compared spatial firing bias between PoS and RSC Correlations of firing rate with linear running speed and AHV by computing the ratio of data to control values for each ses- were computed as described in section ‘Methods’, yielding per- sion and then comparing between cell types for each of the centage firing rate change as a function of movement speed. For three spatial parameters. For spatial information, the ratio of linear running speed, there was a weak correlation of firing rate data: control was 1.63 ± 0.06 for PoS and 2.22 ± 0.10 for RSC, with running speed relative to the control 10 Hz spike train; this −7 which was significantly different (t(145) = 5.0, p = 7 × 10 , was 0.69 ± 0.1%/m/s for PoS and 0.61 ± 0.1%/m/s for RSC. A two- d = 0.83). For coverage, the ratio for PoS was 0.98 ± .01 and for way analysis of variance (ANOVA) of data type (real vs 10 Hz RSC was 0.99 ± .01, which did not differ (t(145) = 0.67, control) and brain area found a main effect of data (F(1, 1) = 40.1, p = 0.25). For coherence, the ratio for PoS was 1.26 ± 0.04, and p < 0.0001), but no effect of brain area (F(1, 1) = 0.50, p = 0.5) and for RSC it was 1.48 ± 0.04, which was significantly different no interaction (F(1, 64) = 0.30, p = 0.56). −4 (t(145) = 3.67, p = 1.7 × 10 , d = 0.61). For AHV, there was also overall a very weak relationship Overall, then, both PoS and RSC HD neurons showed a with firing rate relative to the control steady 10 Hz spike train; small amount of spatiality to their firing when compared with this was 0.12 ± 0.01%/°/s for PoS and 0.08 ± 0.01%/°/s for a time-shifted version of the same data, having higher spatial RSC. These effects, though small, were significant: a two-way information and spatial coherence; this was more pronounced ANOVA comparing data versus control slopes for PoS and in RSC. This accords with the visual inspection showing occa- RSC found a main effect of data type (F(1, 1) = 9.70, p = 0.003), sional and reproducible spatial inhomogeneity of firing. It thus a main effect of brain area (F(1, 1) 5.00, p = 0.03) and a sig- appears that there is a degree of spatial modulation of firing, nificant interaction (F(1, 64) = 4.20, p = 0.025). Plots of firing but this was rather slight. For RSC, this is surprising given rate against HD and AHV (Figure 3) revealed an additional previous reports of conjunctive spatial and directional firing in difference between PoS and RSC neurons. An example of this structure (Cho and Sharp, 2001). However, it is consistent each (the closest cell to the mean in each case) is shown in with our previous studies of RSC HD cells, in which we have Figure 3(a), where it can be seen that the cell’s directional tun- observed very little clear spatial firing (Jacob et al., 2016; ing deviates as a function of AHV, with greater deviation for Knight et al., 2014). higher AHVs in the RSC but not the PoS neuron. This devia- tion reflects anticipatory firing, first reported by Sharp and colleagues for anterior thalamic HD neurons but not PoS Temporal components of firing (Blair et al., 1997; Blair and Sharp, 1995) and replicated by Taube and Muller (1998); anticipatory firing was subse- To analyse temporal patterns of spiking, we took one trial (usually quently shown to a lesser extent for RSC HD neurons the first) for each cell and used the ISI histogram (Supplementary (Cho and Sharp, 2001). It is determined here by taking the Figure 4) to determine the typical time between spikes (ISI peak) slope of the AHV/PFD relationship. Comparing PoS and RSC and spread of firing intervals (decay time to half-peak; Table 1). Lozano et al. 9 Figure 3. Modulation of HD cell firing by angular head velocity (AHV). (a) The heat plots show firing rate (colour; max = red) as a function of HD (y-axis) and AHV (x-axis) for a PoS and RSC neuron. Tuning curves collapsed across all AHVs for left and right head turns are shown to the left and right sides, respectively. Firing rate collapsed across HD as a function AHV is shown below each plot (means and standard error of the mean (SEM) of the shuffled control shown in blue). The left–right downwards slope evident in the RSC plot reveals anticipatory firing, in which the tuning curve shifts in the positive direction (right) for left head turns and in the other direction for right head turns. The shift increased linearly with AHV, revealing a constant time lag. (b) Overall, anticipatory firing differed between PoS and RSC (dotted line shows mean). 10 Brain and Neuroscience Advances individual co-recorded cells in a trial was averaged to yield an overall value. Overall cue control We first looked at whether there was overall cue control by the cue pair, irrespective of the visual stimuli feature, and whether this would differ for brain region or cue type. As described in section ‘Methods’ (Figure 2), the firing directions for each trial in a session were extracted and re-oriented relative to the session mean, so as to express them all in the same reference frame. Then, because visual inspection suggested that firing directions were sometimes distributed in a bipolar fashion, we doubled the angles modulo 360 and re-computed the firing directions; this has the effect of wrapping points at 180° around to 0/360°, and rendering a bipolar distribution unipo- lar. The angle-doubled data were compared against a control, shuffled dataset in which the PFD for each trial was randomly generated; this procedure was repeated until 1000 such pseudo-sessions had been obtained. Overall, the 2140 cells × trials (149 cells in multiple trials) comprised 221 cells × sessions, which collapsed (after averag- ing across cells) to 113 sessions, of which 41 were from PoS and 72 from RSC. Overall, the number of cells recorded in each cue condition were 48, 166 and 17 for HIGH, MOD and LOW cues, respectively. The shuffle control procedure yielded R-vector scores of 0.31 ± 0.01 for both single and doubled val- ues, so this was used as the threshold against which to statisti- cally evaluate cue control. The population of all firing directions before and after angle- doubling are shown in Figure 4(a). R-vector scores showed a marked increase after angle-doubling, from 0.56 to 0.74, indicating Figure 4. Overall cue control. The circular dot plots show the raw cell a degree of bipolarity in the original firing directions, therefore to data, expressed relative to session means (0°); the histograms below evaluate overall cue control as a function of brain area, the angle- these show the same data binned in 6° bins, linearised (centred on doubled values were used. Overall, the angle-doubled R-vector 0°) and expressed as a proportion of the total cell count in order to scores far exceeded the shuffled control threshold (t(112) = 23.00, enable easier visual comparison between the cue types. (a) Angle- p < 0.0001), indicating a high degree of cue-following. Average singled or doubled mean firing directions relative to the cue pair, for per-session R-vector scores for PoS were 0.73 ± 0.04 and for the individual trials. For the original angles, each cell’s preferred firing RSC were 0.79 ± 0.02; these values did not differ (two-tailed direction (PFD) was computed relative to Cue 1, and mean values t(111) = 1.42, p = 0.16), indicating no difference between brain computed for each session, each PFD was then realigned to this session areas in the overall level of cue control (Figure 4(b)). mean (0°). For the doubled angles, the PFD value relative to the cue was doubled, modulo 360°, so as to wrap values near 180° around to Cue discrimination zero and thus remove the bimodality. (b) The doubled-angled values compared between PoS and RSC for all trials. Statistics reported in the We next looked at cue discrimination as a function of cue type, first text were calculated using the session means rather than the set of using the original non-doubled angles. If the cells were confusing individual trials. cues, then the firing should be bipolar and the R-vector scores low. Indeed, firing directions (Figure 5) were more clustered for the more discriminable cues. R-vector scores were much higher for the neurons (Figure 3(b)), we found that, as in previous studies, HIGH cues (0.73 ± 0.05) than for the MOD cues (0.55 ± 0.03) or RSC neurons showed a greater ATI (47.91 ± 4.28 ms) than did the LOW cues (0.27 ± 0.08). Unbalanced one-way ANOVA found PoS neurons (14.37 ± 8.19 ms); these were significantly differ- this difference to be highly significant (F(2, 110) = 8.0, p = 0.0006), ent (one-tailed t(147) = 3.62, p = 0.0002). The PoS values did and post hoc testing (Tukey’s honest significant difference (HSD)) not differ from zero (t(74) = 1.75, p = 0.0836). found that the HIGH cues had significantly higher R-vector scores than both the MOD cues (t(101) = 2.18, p = 0.03) and the LOW cues (t(23) = 6.58, p = 0.0001), while the MOD and LOW cues also Cue control differed (t(96) = 2.90, p = 0.005). The next set of analyses investigated ensemble behaviour and In order to determine whether the differences might be due aimed to determine the extent to which the cells were con- to differential confusion between cues, we repeated the analysis trolled by the visual cues: for these analyses, behaviour of the using the angle-doubled values to remove bipolarity. R-vector Lozano et al. 11 Figure 5. Comparison of responding to the three categories of visual cues. Dot plots and histograms are generated as for Figure 4. (a) Examples of eight recording trials from three HD cells, one from each cue condition, showing the relationship of firing direction (polar plot) to the index cue (black dot). Peak firing in Hz is shown below each plot. The first two cells are from RSC, the last from PoS. The cell in the HIGH discriminability condition showed much less variability, as was typical. (b) and (c) Population data. PoS and RSC data have been combined. (b) The original angles; (c) the doubled angles, showing increased dispersion for the highly discriminable cues (HIGH) but decreased dispersion for the MOD and LOW cues, indicating that their original firing directions had a degree of bipolarity. See text for statistical comparisons. 12 Brain and Neuroscience Advances differences in resulting R-vector score (F(2, 85) = 0.90, p = 0.39) so the data were analysed together. Basic single- angled R-vector scores for the two brain areas did not differ (PoS = 0.57 ± 0.05, RSC = 0.55 ± 0.04, t(85) = 0.23, p = 0.82). Similarly, although the doubled-angle scores significantly improved overall, from 0.55 ± 0.03 to 0.79 ± 0.02 (one-tailed −10 paired t(87) = 7.13, p = 1.37 × 10 ), the doubled-angle Rayleigh scores were not different (t(85) = 0.33, p = 0.74). Thus, there was no difference in the degree of cue control ver- sus cue confusion between brain areas. Comparison of visual and non-visual contributions to cue discrimination For five recording sessions (48 trials) from one rat (the sham LGN-lesioned one), a test was made to see whether the discrimi- nation was purely based on visual pattern, or whether there might have been a contribution from the olfactory/tactile com- ponents of the cue cards. This was done using the V-H and T-B cue cards, which could be reversed in visual identity (V to H, or T to B, and vice versa) by rotating them around their centre points (90° to make V become H, and 180° to make T become B). Data were then analysed with firing aligned in the visual-cue reference frame (where Cue 1 was the V or T cue, irrespective of physical card identity) or in the physical reference frame (where Cue 1 is the same card across trials, regardless of its orientation). The alignment of PFDs is shown in Figure 7. In both reference frames there was a degree of bipolarity, indicating that the PFD sometimes followed the secondary cue in a given reference frame, but cue control was stronger in the visual reference frame (R-vector of 0.54) than in the physical one (R = 0.30). When R-vectors were compared across sessions, there was a decline in the score when the visual-frame data were compared with the physical-frame data (visual = 0.54 ± 0.05; physical = 0.31 ± 0.08). This difference only just reached statistical significance Figure 6. Comparison of responding to the MOD cues. Preferred firing (t(4) = 2.2, one-tailed p = 0.046), but the improvement from sin- directions were derived and plotted, together with R-vector score and gle to double-angle R-scores hints at a contribution from both normalised histograms, as for Figure 5. There was an improvement sensory modalities, with visual being slightly stronger. in score with the doubled-angle values, indicating a degree of cue confusion, but no difference in cue control between PoS and RSC (see RSC and PoS HD cells produced text for details). unipolar tuning curves even when the landmarks were identical scores for the HIGH cues were 0.62 ± 0.06, for the MOD cues were 0.79 ± 0.04 and for the LOW cues were 0.77 ± 0.23. ANOVA showed that scores for the HIGH cues were now sig- No evidence of bipolar tuning curves was seen by visual inspec- nificantly lower than for the MOD cues (t(101) = 3.01, p = 0.003) tion. However, to check quantitatively, an autocorrelation was and not different for the LOW cues (t(23) = 1.63, p = 0.12); nor run on the data (see section ‘Methods’) in order to determine did the MOD and LOW cues differ (t(96) = 0.36, p = 0.72). whether there was a secondary peak at 180° due to periodic Thus, the superior single-angle R-vector scores of the HIGH reversal of the tuning curve arising from cue confusion. Session cues can be attributed to the more bipolar distribution of the data were combined into a single dataset for a given cell, and the MOD and LOW cues. autocorrelation peaks extracted. Identical-cue trials were ana- To compare between brain regions, we then looked in more lysed separately because there were too few of them: a two-way detail at responses to the MOD cues, as these were the ones ANOVA of the remaining conditions found no significant differ- that were in principle discriminable but in practice evoked ences, either as a function of cue type (F(4, 48 = 0.26, p = 0.90), some confusion, suggesting maximal demand on landmark dis- autocorrelation peak (F(1, 12) = 0.59, p = 0.46) or interaction crimination processing. However, we found no difference. The (F(4, 48 = 0.11, p = 0.36). A paired t-test on the identical-cue con- overall single- and doubled-angle data for all the MOD trials dition, where bipolarity might have been strongest if it occurred, are shown in Figure 6, as a function of brain area. Comparison was also non-significant (t(12) = 1.54, p = 0.07). Thus, there was of the three cue types (R-L, T-B and V-H) yielded no no hint of bipolarity in the tuning curves within a single trial. Lozano et al. 13 Figure 7. Reversal of firing provoked by visual change alone. Data from a rat in which cues were changed in visual identity independently of their physical identity, by rotating the cue cards so that a vertical bar became horizontal or a top bar became bottom, etc. When activity was aligned relative to the visual cue, activity was clustered, but slightly bipolar. When aligned to the physical cue identity, firing was more dispersed. When angles were doubled to remove bipolarity (which has the same final effect for both reference frames), the data were highly clustered around zero, indicating overall good control by the cue pair. See text for session statistics. cues (F(2, 228) = 0.2, p = 0.83). The y-intercept measure showed Stabilisation of PFD across a trial a barely significant difference (F(2, 228) = 3,1, p = 0.048), but this The time course of the establishment of firing direction was com- went in the opposite direction from expected (highly discrimina- puted using a stabilisation analysis (see section ‘Methods’), look- ble cues had a higher y-intercept, at 21.14 s) and may have been ing at either percentage of total spikes emitted in each 30-s time due to the low n (n = 17) for the LOW cues. decile, or time taken to achieve each spike count decile (Figure 8). In summary, the stabilisation analysis revealed a slight ten- The percentage of spikes emitted in the first 30 s was 8.27 ± 0.13%, dency for a slower accumulation of spikes (in other words, slower which was significantly less than the expected 10% (t(230) = 13.11, development of firing), especially in the PFD, in the first 20–30 s p < 0.0001). The z-score for this decile was −0.78, while for the of each trial, but this did not depend on brain region or cue type. remaining deciles it did not deviate more than 0.44 from zero Overall, this analysis reveals that actually there was a propensity (Figure 8(a)). A corresponding observation emerged from the for firing to become established very soon after entry into the y-intercept analysis. The plots in Figure 8(b) are highly linear, with environment, even for cues that were harder to discriminate. R values close to 1 for most sessions (0.9979 ± 0.0002 for the 231 cells × sessions overall). However, the regression lines did not go through zero: the y-intercepts averaged 18.22 +/- 1.05 s, which was Discussion significantly different from zero (t(230) = 21.44, p < 0.0001). Thus, This study compared the activity of PoS and RSC HD cells in a the time taken to accumulate the first decile’s worth of spikes in the landmark discrimination setting, in which two opposing land- PFD range was almost 20 s longer than would be expected based marks had similar overall shape and/or contrast/luminance but on the subsequent deciles. In order to determine whether this effect differed in their fine-grained visual structure. The study was was due to dispersion of firing around multiple directions or just to motivated by previous findings that lesions to the PoS or RSC a general decline in spiking, we repeated the analysis using all impair both directional tuning stability and landmark control of spikes, not just the ones in the PFD range; a similar, albeit slightly HD cells in other parts of the circuit, suggesting a role for these attenuated effect was evident (Figure 8), suggesting a general structures in directional landmark processing. The aims were reduction of activity, rather than spatial dispersion, in the first 20– twofold: (a) to see whether the two regions differed in their 30 s of environment exploration. We looked at movement corre- responsiveness, which might provide clues to their function, and lates including distribution of HD with respect to the PFD, linear (b) to see whether HD cells could distinguish the landmarks by velocity and AHV, and did not find any difference between the first maintaining a consistent directional relationship to the array, and decile and subsequent ones that could explain this observation (not if so, to determine how quickly this discrimination occurred, and shown). This slight reduction in spiking suggests that on entry into whether there were regional differences. These two considera- an environment there is a period of time during which the system tions are discussed in turn. is acquiring the information needed to drive the cells to firing threshold. We compared these measures between PoS and RSC but Firing properties of PoS versus RSC HD found no differences. The total percentage of spikes accumulated neurons in the first 30 s of the trial did not differ between PoS and RSC The basic firing properties we saw were similar to those reported (t(229) = 1.86, p = 0.06) and the y-intercept measure was also not previously, with the exception that the peak firing rates we observed different (t(229) = 1.87, p = 0.06). We finally looked at the two in PoS were considerably lower than those that have been reported measures as a function of cue type. The spike count percentages previously: ~8 Hz in our study compared with ~35 Hz reported by in the first 30 s did not differ between HIGH, MOD and LOW 14 Brain and Neuroscience Advances Figure 8. Stabilisation of the preferred firing direction (PFD) on environment entry. Data from a 40° range surrounding the PFD were divided into deciles of either time (30 s bins) or spikes (10% total). (a) Top: Spike count as a function of time decile revealed a reduced spike count in the first 30 s (circled). Bottom: z-score analysis of those data revealed a larger negative score for the first decile relative to the others, present for both spikes in the PFD range and to a lesser extent in all the spikes. See Results for statistical tests. (b) Top: The spikes from each 300 s trial were divided into 10 deciles, and the time to reach each decile plotted. Each fine grey line represents the plot from one cell over one session (averaged across trials); the solid black line is the mean across all cells/sessions and the dotted line extrapolates to the y-axis. Bottom: the y-intercepts, which should be at zero, were clustered around a mean (black line) of 18 s for the within-PFD spikes, indicating that it took extra time to reach the first decile relative to subsequent deciles. As with the spike count data, the effect was attenuated but still present when all spikes were considered. Taube et al. (1990b) and Taube and Muller (1998) and 19 Hz for The observation that RSC neurons anticipate upcoming HD’s, Sharp (1996). The reason for the large discrepancies between stud- in our case by 48 ms, is similar to that published previously. ies might be due to the recording methods used, and the possible Anticipatory firing was first reported by Blair and Sharp (1995) difficulty of the single-wire recording methods of earlier studies to for anterior thalamic HD neurons, which were found to predict identify low-firing-rate neurons. In support of this, a more recent future HD by around 37 ms; this contrasted with PoS which did study using tetrodes, as in the present study, found a rate of 3 Hz in not show anticipation. Taube and Muller (1998) reported a value PoS HD cells (Brandon et al., 2012). We also averaged across sev- of around 23 ms for anterior thalamus, and −7 ms (firing lagging eral trials, which may have brought the rates towards the mean. HD) for PoS HD neurons. In a later study, RSC HD neurons were Relatively few differences in firing properties were seen in HD also found to show anticipation, of around 25 ms (Cho and Sharp, cells from the two brain regions, but those that we observed repli- 2001). In this study, we also saw no anticipation in PoS neurons. cated previous observations: (a) firing rates were higher in RSC; The implication is that RSC firing very slightly precedes PoS fir- (b) tuning curves were slightly broader and (c) RSC firing antici- ing; it may be, therefore, that the route of movement-related pated HD by about 48 ms. The observation of higher firing rates in information flow is from RSC to PoS. This is consistent with the RSC is consistent with a report by Sharp (2005), who found similar observation that granular RSC neurons project to layer III PoS but smaller differences in firing rate between these regions. We (Kononenko and Witter, 2012), which is where many HD cells in also found that firing rate was weakly (albeit significantly) corre- this region are found (Boccara et al., 2010; Preston-Ferrer et al., lated with running speed, with no differences between PoS and 2016), the remainder being in the deep layers (Boccara et al., RSC. However, PoS was slightly more influenced by AHV 2010; Preston-Ferrer et al., 2016; Ranck, 1984; Taube et al., although modulation was low for both regions. 1990b). Anticipation may reflect vestibular or motor efference Lozano et al. 15 copy signals, providing advance warning of upcoming head as proposed both by attractor network models (Knierim and directions which can then be combined with descending land- Zhang, 2012) and following population recordings of HD cells mark-related sensory inputs. A study of ADN neurons found that (Peyrache et al., 2015; Seelig and Jayaraman, 2015). anticipatory firing also occurred, and indeed more prominently, Having established cue control by the landmark pair, we in association with passive head movements (Bassett et al., then looked to see whether the cues within the pair could be 2005), suggesting a stronger vestibular contribution (Van der further discriminated. We found that discrimination of the Meer et al., 2007). Clark et al. (2010) found that neurotoxic RSC non-identical cues was highest for the high-contrast HIGH lesions increase anticipatory firing in the ADN, possibly due to (black/white) cue pair but was also well above control levels removal of an environment-anchoring stimulus and consequent for the patterned, moderately discriminable (MOD) cues too. overweighting of the vestibular AHV inputs. The precise contri- For one rat, we dissociated visual and olfactory characteristics butions of the various self-motion signals, including how they of the cue cards and found a clear contribution from vision. may be differentially weighted in different species and/or under These data show that these cortical HD regions, which are one different conditions, remains to be determined (Cullen, 2014). synapse away from the visual cortex (Vogt and Miller, 1983; Because of the connections of both regions with the hip- Van Groen and Wyss, 1992), can access information about pocampal formation, and because some theoretical models of HD internal structure of landmarks – furthermore, discrimination cell processing require spatial inputs (Bicanski and Burgess, occurs almost immediately upon entry into the environment, 2016), we looked for spatial correlates of firing in the form of and remains stable thereafter. place fields, or at least of reliable spatial heterogeneity. Both When we looked at the period of time immediately after envi- brain regions yielded some cells showing weak spatial encoding, ronment entry we found a gradual increase in firing rate for the but in general spatial modulation was low. For PoS, this observa- first few seconds, reflected in a longer time to accumulate the first tion is consistent with previous studies from this area in which decile’s worth of spikes relative to subsequent deciles. Thus, it HD cells typically have neither place nor theta-frequency modu- seems that landmarks do not purely cause directional clustering of lation (Sharp, 1996). However, Cacucci et al. (2004) and Boccara a randomly distributed base level of activity, but add drive to the et al. (2010) did report conjunctive encoding in regions of pre- system over and above baseline. Two previous studies have subiculum very close to PoS – it is thus likely that there are looked at the dynamics of cue control and have found that regional differences in the occurrence of the difference cell types. although it takes a significant amount of time – minutes – for a Overall, however, it seems unlikely that spatial inputs provide a novel cue to gain control of HD cells (Goodridge et al., 1998), strong input to the HD cell signal directly (although the possibil- cells respond to rotation of a familiar cue within a few tens of mil- ity exists that spatial inputs gate landmark inputs and thus exert liseconds (Zugaro et al., 2003). Our situation is different because their influence in a more subtle manner). although the landmarks were generally familiar, measurements were made within the first few seconds of entry into the environ- ment, and also the cues needed to be discriminated. This could be Landmark discrimination due to the slow onset of intrinsic network stabilisation, or it may One purpose of this study was to examine possible differences in be that a number of cognitive factors come into play that influence landmark processing by the two regions. Cells from both regions HD cell firing in these first few seconds – the animal may need in all the different cue conditions had unipolar tuning curves that time to adjust to the situation and start attending to spatial cues, were clustered around a single direction relative to the cue pair, there may be a lag while the cues are identified and the discrimi- indicating that they could distinguish the cues from the back- nation is made and so on. It would be interesting to look at firing ground and reliably orient by them. This was true even for the rate if the animal were placed into the arena without the cues, to visually identical cues; there was no hint that within a trial, the try and tease apart these possibilities. cells flipped their firing from one direction to its polar opposite. The rapidity with which landmark discrimination was There are two possible reasons for this – one is that the rats could established, in the absence of explicit training, suggests that still distinguish the cues by non-visual means (e.g. perhaps by this two-cue HD cell paradigm may be a good experimental olfaction) and did in fact fire in a unipolar fashion with respect to method with which to test visual discrimination capabilities of the cues. This seems unlikely since there was confusion between rats: for example, following lesions to visual brain areas. cues even for the ones that were moderately discriminable; also, Because explicit training may bias animals to focus on single for the one rat in which the cues were identified by the experi- features of multidimensional stimuli (Jeffery, 2007), sponta- menter and visual and physical identities switched between trials, neous recognition is preferable, but such processes may be the firing directions were less well predicted by physical identity. hard to detect in behavioural output. Cell activity on the other The other, more likely reason is that firing was stabilised within a hand is directly observable and easily measured. The fact that trial by self-motion signals. In other words, having initially HD cells rapidly detect and discriminate complex stimuli guessed at one of the two possible orientations of the cue pair on means that they may serve as a useful read-out for sensory entry into the environment, the cells were then anchored to that perception occurring within seconds or even faster. orientation with the help of self-motion cues (since the rat knew, Overall, therefore, we found relatively few differences from these cues, that it had not suddenly reversed direction). between PoS and RSC in their cue responsiveness. The similarity Henceforth, the system could use the ambiguous landmark pair in responding can be explained either by similar computations together with the self-motion cues to generate a stable directional being performed on the incoming visual signals by the two regions signal, whereby the landmarks would prevent the signal drifting, or else by rapid communication between regions. It seems a priori and the self-motion cues would prevent it from flipping. This unlikely that the regions perform the same computations, but to proposition is consistent with the operation of attractor dynamics, address this question in more detail it will be necessary to isolate 16 Brain and Neuroscience Advances Cacucci F, Lever C, Wills TJ, et al. (2004) Theta-modulated place-by- the regions, either by lesioning each in turn or by targeted opto- or direction cells in the hippocampal formation in the rat. Journal of chemogenetic interventions that functionally disconnect them. Neuroscience 24(38): 8265–8277. This study additionally introduces a new spontaneous visual dis- Calton JL, Stackman RW, Goodridge JP, et al. (2003) Hippocampal crimination method for probing visual processing by the spatial place cell instability after lesions of the head direction cell network. system and reveals a basic capacity for the two cortical HD cell Journal of Neuroscience 23(30): 9719–9731. areas to rapidly detect and discriminate landmarks. These experi- Calton JL, Turner CS, Cyrenne D-LM, Lee BR, Taube JS (2008) Land- ments open the door to investigations of transformation of the mark control and updating of self-movement cues are largely main- visual pathway following lesions or selective inactivation of tained in head direction cells after lesions of the posterior parietal inputs to these areas. Our prediction is that even when the object cortex. Behavioral Neuroscience 122(4): 827–840. processing regions of the brain, such as perirhinal cortex, are Cho J and Sharp PE (2001) Head direction, place, and movement corre- lates for cells in the rat retrosplenial cortex. Behavioral Neuroscience removed, HD neurons will still be able, using basic visual inputs 115(1): 3–25. from primary visual areas, to discriminate similar landmarks and Clark BJ, Bassett JP, Wang SS, et al. (2010) Impaired head direction cell use them for orientation. representation in the anterodorsal thalamus after lesions of the retro- splenial cortex. Journal of Neuroscience 30(15): 5289–5302. Acknowledgements Cullen KE (2014) The neural encoding of self-generated and externally applied movement: Implications for the perception of self-motion The authors are grateful to Miguel Valencia for advice on analysis and and spatial memory. Frontiers Integrative Neuroscience 7: 108. Josh Bassett for useful comments on the article. Golob EJ, Stackman RW, Wong AC, et al. (2001) On the behavioral significance of head direction cells: Neural and behavioral dynam- ics during spatial memory tasks. Behavioral Neuroscience 115(2): Data sharing 285–304. Data supporting the findings of this study are available on request from Goodridge JP and Taube JS (1997) Interaction between the postsubicu- the corresponding author. lum and anterior thalamus in the generation of head direction cell activity. Journal of Neuroscience 17(23): 9315–9330. Goodridge JP, Dudchenko P, Worboys K, et al. (1998) Cue control and Declaration of conflicting interests head direction cells. Behavioral Neuroscience 112(4): 749–761. KJ is a non-shareholding director of Axona Ltd. Jacob PY, Casali G, Spieser L, et al. (2016) An independent, landmark- dominated head direction signal in dysgranular retrosplenial cortex. Nature Neuroscience 20(2): 173–175. Funding Jankowski MM, Islam MN, Wright NF, et al. (2014) Nucleus reuniens of The work was supported by funding from the UK Medical Research the thalamus contains head direction cells. eLife 3: e03075. Council (G1100669) and the Wellcome Trust (WT103896AIA) to KJ Jeffery KJ (2007) Do discrimination tasks discourage multi-dimensional and a studentship from the Mexican Consejo Nacional de Ciencia y stimulus processing? Evidence from a cross-modal object discrimi- Tecnologia (3132554) to YL. nation in rats. Behavioural Brain Research 183(2): 213–221. Kadir SN, Goodman DFM and Harris KD (2014) High-dimensional clus- ter analysis with the masked EM algorithm. Neural Computation References 26(11): 2379–2394. Abeles M (1982) Quantification, smoothing, and confidence limits for Knierim JJ and Zhang K (2012) Attractor dynamics of spatially corre- single-units’ histograms. Journal of Neuroscience Methods 5(4): lated neural activity in the limbic system. Annual Review of Neuro- 317–325. science 35: 267–285. Bassett JP and Taube JS (2001) Neural correlates for angular head veloc- Knight R, Hayman R, Ginzberg LL, et al. (2011) Geometric cues influ- ity in the rat dorsal tegmental nucleus. Journal of Neuroscience ence head direction cells only weakly in non-disoriented rats. Jour- 21(15): 5740–5751. nal of Neuroscience 31(44): 15681–15692. Bassett JP, Zugaro B, Muir GM, et al. (2005) Passive movements of the Knight R, Piette CE, Page H, et al. (2014) Weighted cue integration in head do not abolish anticipatory firing properties of head direction the rodent head direction system. Philosophical Transaction of the cells. Journal of Neurophysiology 93(3): 1304–1316. Royal Society of London, Series B: Biological Science 369(1635): Berens P (2009) CircStat: A MATLAB toolbox for circular statistics. Journal of Statistical Software 31(10): 1–21. Kononenko NL and Witter MP (2012) Presubiculum layer III conveys Bicanski A and Burgess N (2016) Environmental anchoring of head retrosplenial input to the medial entorhinal cortex. Hippocampus direction in a computational model of retrosplenial cortex. Journal 22(4): 881–895. of Neuroscience 36(46): 11601–11618. Paxinos G and Watson C (2007) The Rat Brain in Stereotaxic Coordi- Blair HT and Sharp PE (1995) Anticipatory head direction signals in nates (6th edn). Cambridge, MA: Elsevier Academic Press, pp. anterior thalamus: Evidence for a thalamocortical circuit that inte- 547–612. grates angular head motion to compute head direction. Journal of Peyrache A, Lacroix MM, Petersen PC, et al. (2015) Internally organized Neuroscience 15(9): 6260–6270. mechanisms of the head direction sense. Nature Neuroscience 18(4): Blair HT, Lipscomb BW and Sharp PE (1997) Anticipatory time intervals 569–575. of head-direction cells in the anterior thalamus of the rat: Implica- Peyrache A, Roux L, Schieferstein N, et al. (2016) Transformation of tions for path integration in the head-direction circuit. Journal of head-direction signal into spatial code. bioRxiv. Epub ahead of print Neurophysiology 78(1): 145–159. 20 September. DOI: 10.1101/075986. Boccara CN, Sargolini F, Thoresen VH, et al. (2010) Grid cells in pre- Preston-Ferrer P, Coletta S, Frey M, et al. (2016) Anatomical organization and parasubiculum. Nature Neuroscience 13(8): 987–994. of presubicular head-direction circuits. eLife 5: e14592. Brandon MP, Bogaard AR, Andrews CM, et al. (2012) Head direc- Ranck JB (1984) Head-direction cells in the deep layers of the dor- tion cells in the postsubiculum do not show replay of prior waking sal presubiculum in freely moving rats. Society for Neuroscience sequences during sleep. Hippocampus 22(3): 604–618. 10: 599. Lozano et al. 17 Seelig JD and Jayaraman V (2015) Neural dynamics for landmark ori- Taube JS, Muller RU and Ranck JB Jr (1990a) Head-direction cells entation and angular path integration. Nature 521(7551): 186–191. recorded from the postsubiculum in freely moving rats I: Descrip- Sharp PE (1996) Multiple spatial/behavioral correlates for cells in the rat tion and quantitative analysis. Journal of Neuroscience 10(2): postsubiculum: Multiple regression analysis and comparison to other 420–435. hippocampal areas. Cerebral Cortex 6(2): 238–259. Taube JS, Muller RU and Ranck JB Jr (1990b) Head-direction cells Sharp PE (2005) Regional distribution and variation in the firing proper- recorded from the postsubiculum in freely moving rats II: Effects ties of head direction cells. In: Wiener SI and Taube JS (eds) Head of environmental manipulations. Journal of Neuroscience 10(2): Direction Cells and the Neural Mechanisms of Spatial Orientation. 436–447. Cambridge, MA: MIT Press, pp. 3–15. Van der Meer M, Knierim JJ, Yoganarasimha D, et al. (2007) Anticipa- Skaggs WE, Gothard KM and Markus EJ (1993) An information-the- tion in the rodent head direction system can be explained by an inter- oretic approach to deciphering the hippocampal code. In: Hanson action of head movements and vestibular firing properties. Journal SJ, Cowan JD and Glies CL (eds) Advances in Neural Information of Neurophysiology 98(4): 1883–1897. Process Systems 5. San Francisco, CA: Morgan Kaufmann Publish- Van Groen T and Wyss JM (1992) Connections of the retrosplenial dys- ers, pp. 1031–1037. granular cortex in the rat. Journal of Comparative Neurology 315(2): Taube JS (2007) The head direction signal: Origins and sensory-motor 200–216. integration. Annual Review of Neuroscience 30(1): 181–207. Vogt BA and Miller MW (1983) Cortical connections between rat cingu- Song P and Wang X-J (2005) Angular path integration by moving “hill of late cortex and visual, motor, and postsubicular cortices. Journal of activity”: A spiking neuron model without recurrent excitation of the Comparative Neurology 216(2): 192–210. head-direction system. Journal of Neuroscience 25(4): 1002–1014. Yoder RM, Clark BJ and Taube JS (2011) Origins of landmark encoding Taube JS and Burton HL (1995) Head direction cell activity monitored in in the brain. Trends in Neurosciences 34(11): 561–571. a novel environment and during a cue conflict situation. Journal of Yoder RM, Peck JR and Taube JS (2015) Visual landmark information Neurophysiology 74(5): 1953–1971. gains control of the head direction signal at the lateral mammillary Taube JS and Muller RU (1998) Comparisons of head direction cell nuclei. Journal of Neuroscience 35(4): 1354–1367. activity in the postsubiculum and anterior thalamus of freely moving Zugaro MB, Arleo A, Berthoz A, et al. (2003) Rapid spatial reorientation rats. Hippocampus 8(2): 87–108. and head direction cells. Journal of Neuroscience 23(8): 3478–3482.

Journal

Brain and Neuroscience Advances – SAGE

Published: Sep 15, 2017

Keywords: Head direction cells; postsubiculum; retrosplenial cortex; landmarks; visual discrimination; spatial memory; in vivo rodent electrophysiology

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Retrosplenial and postsubicular head direction cells compared during visual landmark discrimination:

Retrosplenial and postsubicular head direction cells compared during visual landmark discrimination:

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Retrosplenial and postsubicular head direction cells compared during visual landmark discrimination:

Retrosplenial and postsubicular head direction cells compared during visual landmark discrimination:

References (99)

Abstract

Journal

Recommended Articles

There are no references for this article.

Our policy towards the use of cookies