As you enter a new environment such as visiting a classroom for the first time, your brain takes in information about your surroundings to help inform where you are and what direction you are facing.
Knowing where the center of the room is located helps provide a reference point for processing space.
A Dartmouth study published in Science provides new insight into navigation and spatial learning by examining how the rat brain processes spatial information from the outside world in an egocentric framework and converts it to information in relation to the animal’s spatial position in an allocentric framework (referenced to the world at-large).
As the study explains, the postrhinal cortex of the rat brain is considered “the rodent homolog of the human parahippocampal cortex,” which is thought to be responsible for spatial navigation.
The study examined how the rat brain processes incoming sensory information to inform where it is. For each trial, the rat was placed in an open box (“room”) and trained to forage for sugar pellets that fell from overhead.
The rat was filmed and its brain activity was recorded using electrodes.
Following the trials, the researchers analyzed what the spatial cells were doing in relation to where the rat was in the environment. The results demonstrated that the postrhinal cortex contains three types of spatial cells, which act together to provide a sense of where the rat is located and its directional orientation within the local environment.
- Some neurons are “center-bearing” cells, which are egocentric in nature, and identify where the geometric center of the space is located in relation to the animal. The center-bearing cells inform the animal of its directional orientation relative to where the center of room is, such as whether the room center is in front or behind the animal.
- Other neurons are “center-distance” cells, whose firing rates encode the distance the rat is from the center of the environment. Some center-distance cells may fire more rapidly in the center while others may fire more slowly there, but both types of these cells are encoding the same type of information.
- Finally, other neurons are “head-direction” cells, which respond to and are specific to a cardinal direction. Regardless of where the animal is located in the room, the cell will fire whenever the animal is looking in a specific direction. Different head direction cells encode different directions and the population of this cell type appears to encode all 360 degrees.
Together, the center-bearing cells, center-distance cells and head-direction cells tell the animal where it is and what direction it is facing.
“Our results demonstrate that it appears that the rat uses the center of the environment to ground where its located, providing new insight into how these animals locate and orient themselves in an area that has boundaries,” explains lead author Patrick A. LaChance, a graduate student in the department of psychological and brain sciences at Dartmouth.
“In the area of spatial learning, the study’s findings are exciting, as they illustrate how the postrhinal cortex could provide a template for how humans convert egocentric information (the reference frame in which all information enters the brain) into allocentric information (the reference frame relative to the external world) in the parahippocampal cortex, says co-author Jeffrey S. Taube, a professor of psychological and brain sciences at Dartmouth, whose lab focuses on spatial cognition and the neural correlates of navigation.
Other studies have shown that when humans experience brain damage to the parahippocampal cortex region, they may have topographical disorientation and may not be able to form new spatial representations and often are disoriented and get lost. In sum, the study sheds light on how and where in the brain spatial information gets processed in order to enable us to do things such as find our car in the parking lot at the end of the day and direct our route home.
When faced with novelty, most mammals increase their exploratory behavior to facilitate establishing and refining the internal representation of the novel element that could be for instance a change in the surroundings, a novel object in familiar settings or a conspecific.
As a special case, animals engage in active exploration when placed in a novel environment to gain spatial knowledge and represent the physical landscape.
In the past century, after a series of seminal experiments on rats navigating in a complex maze, Edward Tolman put forward the proposition that animals, including humans, can acquire large numbers of sensory cues and use them to build a mental image of the environment.
With this internal representation of physical space they can navigate to a goal by knowing where it is embedded in a complex set of environmental features (Tolman, 1948; Tolman & Honzik, 1930a,1930b).
This idea is known today as the cognitive map, an abstract ensemble of environmental relationships and paths which determines the possible routes of actions.
Although generally discussed in the context of physical space, the concept of the cognitive map can be generalised to representations of abstract spaces as well, such as hierarchy or time‐series (Epstein, Patai, Julian, & Spiers, 2017).
Navigation can be defined as the process of determining and maintaining a course or trajectory from one place to another (Gallistel, 1990).
In cognitive neuroscience navigation is regarded as a complex goal‐directed behavior that involves processing a variety of sensory and proprioceptive stimuli, storage and recall of information and elaboration of plans.
There are many ways of finding a goal location, from simple direct approach to an easy‐to‐locate proximal target (local navigation) to using well‐learned routes and cognitive maps (wayfinding).
In the simplest case (target approaching or taxis), the goal is directly detectable and the agent only needs to orient towards the observable goal or landmark nearby and then maintain this direction (beacon, O’Keefe & Nadel, 1978).
In more complex situations when the goal is not visible, it might still be possible to locate it using visible distal cues, which can then be used to compute the goal direction (guidance, Morris, 1981).
To reach a destination outside the local environment, the subject can recongise and approach sequential landmarks or proximal places step‐by‐step (recognition‐triggered response or direction, Mallot & Gillner, 2000) or concatenate multiple recognition‐triggered responses in a route (route following or topological navigation).
Notably, a route is a rigid construct that do not involve the creation of a new path.
The latter is only possible using a map‐based strategy, which capitalises on the relational organization that is characteristic to the cognitive map.
The wanderer has access to two kinds of cues: those generated by proprioception including vestibular feedback (idiothetic) and those emerging from the environment such as landmarks (allothetic).
Similarly, spatial information can be represented both by subject‐based coordinates in a self‐centered (egocentric) frame or by world‐based coordinates in a world‐centered (allocentric) frame (Committeri et al., 2004; Galati, Pelle, Berthoz, & Committeri, 2010; Klatzky, 1998, but see also Meilinger & Vosgerau, 2010; Filimon, 2015).
In natural conditions, idiothetic and allothetic signals can be combined to optimise navigation (Poucet et al., 2013; Sjolund, Kelly, & McNamara, 2018), but may be weighed differentially based on their perceived reliability (Chen, McNamara, Kelly, & Wolbers, 2017).
Thus, the selection of appropriate navigational strategies is primarily determined by the perception of space, that is, by the nature of the cues that can be used for navigation, modified by the subjects’ individual predispositions and expectations (Ishikawa & Montello, 2006; McIntyre, Marriott, & Gold, 2003).
Given the complexity of the process and the different ways by which navigation can be implemented, it is unsurprising that neuroimaging and lesion studies have identified an intricate network of structures involved, including the hippocampus, entorhinal cortex, parahippocampal gyrus, medial and right inferior parietal cortex, regions within prefrontal cortex, cerebellum, parts of the basal ganglia, posterior cingulate cortex and retrosplenial cortex (RSC; Guterstam, Björnsdotter, Gentile, & Ehrsson, 2015; Iaria, Chen, Guariglia, Ptito, & Petrides, 2007; Ito, Zhang, Witter, Moser, & Moser, 2015; Maguire et al., 1998; Rochefort, Lefort, & Rondi‐Reig, 2013).
Hereafter, we will direct our focus on the differential role of the hippocampus, the posterior parietal cortex (PPC) and the RSC in spatial navigation, together which areas seem to form a network that may serve as the anatomical bases of flexible integration of allocentric and egocentric information.
Next, we will review the structural properties of cholinergic basal forebrain (BF) afferents to these cortical structures originating from the medial septum (MS) and the nucleus basalis magnocellularis (NBM; Figure 1).
Finally, we will summarize the functional evidence provided by lesion, microdialysis and pharmacology studies on the role of these BF to cortex pathways in controlling different aspects of spatial cognition.
Although discussing other brain areas are beyond the scope of this review, this does not diminish their relevance for navigation (see for example (Mizumori, Puryear, & Martig, 2009; Chersi & Burgess, 2015) on the importance of striatal circuits in spatial navigation).
Anatomical location of the spatial navigation network. (a) 3D, (b) coronal and (c) sagittal views of the mouse brain highlighting the locations of medial septum (red), nucleus basalis magnocellularis (yellow), hippocampus (green), posterior parietal cortex (blue) and retrosplenial cortex (teal). Image credit: Allen Institute
The hippocampus
The medial temporal lobe is a key structure in the spatial domain of cognition, for navigation as well as encoding and retrieval of spatial memory.
This includes the anatomical areas of the hippocampal region (hippocampus proper, dentate gyrus and subicular complex) and the adjacent cortex (perirhinal, entorhinal, and parahippocampal cortices; Aggleton, 2012; Fernández‐Ruiz & Oliva, 2016; Lavenex & Amaral, 2000; Squire & Zola‐Morgan, 1991).
The literature in this field is vast and multidisciplinary.
Here we only provide an overview pertinent to our focus of cholinergic modulation of hippocampal function in the context of spatial learning and memory; for more extensive summaries and thought‐provoking reading please refer to (Eichenbaum et al., 2016; Lisman et al., 2017; Moser, Moser, & McNaughton, 2017). As stated by (O’Keefe & Nadel, 1978), “we shall argue that the hippocampus is the core of a neural memory system providing an objective spatial framework within which the items and events of an organism’s experience are located and interrelated”.
Hippocampus indeed seems to represent spatiotemporally coincident elements of a context, creating allocentric cognitive maps which are then used to guide exploration, plan navigation and interpret the current state of the world (Schiller et al., 2015).
Tasks requiring either spatial memory or navigation based on allocentric cues generally engage the hippocampus (Hartley, Maguire, Spiers, & Burgess, 2003; Iaria, Petrides, Dagher, Pike, & Bohbot, 2003; Kumaran & Maguire, 2005; Maguire et al., 1998) and are impaired in patients with hippocampal brain damage (Corkin, Amaral, González, Johnson, & Hyman, 1997; Feigenbaum & Morris, 2004; Guderian et al., 2015; Hartley et al., 2007; Holdstock et al., 2000; Scoville & Milner, 1957). Spatial navigation becomes gradually impaired during ageing.
A recent study (Konishi, Mckenzie, Etchamendy, Roy, & Bohbot, 2017) found that the age‐related impairment of spatial memory was correlated with decreased hippocampal volume, while better general cognitive functions were associated with superior wayfinding abilities and increased use of hippocampus‐dependent spatial strategies.
A specific loss of cholinergic innervation of the hippocampal formation has been hypothesized as a leading cause for age‐related memory decline both in normal ageing and Alzheimer’s patients (Arendt & Bigl, 1986; Gallagher & Colombo, 1995; Whitehouse, Price, et al., 1982).
It was established that the extent of cognitive impairment was correlated with loss of BF cholinergic neurons from post‐mortem human brain tissue samples (Arendt & Bigl, 1986; Bowen, Smith, White, & Davison, 1976; Iraizoz, de Lacalle, & Gonzalo, 1991; Perry et al., 1978).
Moreover, functional alterations of cholinergic activity might precede morphological degeneration (Palop & Mucke, 2016; Schliebs & Arendt, 2011); however, a direct support for specific cholinergic dysfunction before axonal degeneration and plaque formation is lacking.
Given the well‐known age‐associated impairment of the medial septal cholinergic neurons (Schliebs & Arendt, 2011), this raises the possible causal involvement of diminishing cholinergic innervation in impaired hippocampal spatial memory during ageing.
Although the hippocampus is certainly fundamental for allocentric navigation, it has recently been proposed that such tasks might be implemented by a broader network (Ekstrom, Arnold, & Iaria, 2014).
Indeed, analysing the trajectory of amnestic patients with medial temporal lobe damage in a virtual analogue of the Morris water maze (MWM), it was found that they retained the ability to acquire and utilise coarse spatial maps, with partial allocentric memory (Kolarik, Baer, Shahlaie, Yonelinas, & Ekstrom, 2018; Kolarik et al., 2016).
In rodents, lesions to the hippocampus impairs navigational tasks that require the processing of environmental cues or exploration of new locations, while simple stimulus‐response learning remains intact (Cohen, LaRòche, & Beharry, 1971; Morris, Garrud, Rawlins, & O’Keefe, 1982; Olton, Walker, Gage, & Johnson, 1977). However, the exact nature of the impairment and whether it is the same across rodents and humans remains less clear. Clark and colleagues used a version of the MWM that allowed the animal to proceed directly to the hidden platform using a beacon (Clark, Broadbent, & Squire, 2007).
However, the rats first had to select the correct beacon from four identical objects based on distal cues. After the lesion, animals were unable to reach the platform, as expected if the hippocampus is necessary to process distal cues.
However, contrary to the expectation that rats would look for the platform near the beacons, they did not show an indication of maintaining this concept, suggesting a broader memory impairment not restricted to spatial information.
Additionally, while humans with hippocampal lesions retain spatial memories acquired remotely showing a temporally graded retrograde amnesia, lesioned animals are also impaired on recently acquired memories despite extensive training (Clark, Broadbent, & Squire, 2005).
This could be resolved by raising and training the animals in an enriched environment, where they could acquire an allocentric spatial representation that survived hippocampal damage, confirming that extensive premorbid experience leads to spatial representations that are independent of the hippocampus (Winocur, Moscovitch, Fogel, Rosenbaum, & Sekeres, 2005).
These spatial representations, however, became more schematic and could not support flexible navigation, e.g. choosing an alternative route in the presence of an unexpected obstacle (Winocur, Moscovitch, Rosenbaum, & Sekeres, 2010), similarly to human patients in a complex environment (Maguire, Nannery, & Spiers, 2006).
Crucial support for the navigational role of the hippocampal formation has come from electrophysiology experiments.
Our current understanding of how the brain encodes spatial information has been shaped by the discovery of rodent place cells (O’Keefe & Dostrovsky, 1971), head direction cells (Taube, Muller, & Ranck, 1990), grid cells (Hafting, Fyhn, Molden, Moser, & Moser, 2005), conjunctive grid‐head direction cells (Sargolini et al., 2006), border cells (Solstad et al., 2008) and most recently speed cells (Kropff, Carmichael, Moser, & Moser, 2015), while other specific representations may be present (e.g. see Diehl, Hon, Leutgeb, & Leutgeb, 2017). In the last decade, analogous representations has been discovered in humans (Doeller, Barry, & Burgess, 2010; Ekstrom et al., 2003; Julian, Keinath, Frazzetta, & Epstein, 2018; Killian, Jutras, & Buffalo, 2012; Lee et al., 2017; Miller et al., 2013; Nadasdy et al., 2017; Nau, Navarro Schröder, Bellmund, & Doeller, 2018).
Place cells are hippocampal principal neurons that preferentially fire at given locations of the environment (‘place field’), thus encoding spatial information that collectively allow the reconstruction of the exact location of the animal (Chen, Andermann, Keck, Xu, & Ziv, 2013; O’Keefe & Dostrovsky, 1971; Olypher, Lánský, Muller, & Fenton, 2003; Skaggs, McNaughton, & Gothard, 1993).
Moving across space, place cells are activated in a sequence representing a path that may be reactivated during sleep for memory consolidation (de Lavilléon, Lacroix, Rondi‐Reig, & Benchenane, 2015; O’Neill, Senior, Allen, Huxter, & Csicsvari, 2008; van de Ven, Trouche, McNamara, Allen, & Dupret, 2016) or during wakefulness to recall environmental features and plan actions (Ólafsdóttir, Carpenter, & Barry, 2017; Pfeiffer & Foster, 2013; van der Meer, Johnson, Schmitzer‐Torbert, & Redish, 2010; Wu, Haggerty, Kemere, & Ji, 2017).
This organisation appears to be the default processing scheme of the hippocampus that extends beyond spatial navigation and creates sequential representations of non‐spatial features, probably serving memory‐guided behaviour in general (Allen, Salz, McKenzie, & Fortin, 2016; Aronov, Nevers, & Tank, 2017; Pastalkova, Itskov, Amarasingham, & Buzsaki, 2008). Distal visual and non‐vestibular self‐motion cues can provide enough spatial information to create place fields as observed in virtual reality (VR), but in natural environments where vestibular and other sensory cues are also present, a more robust hippocampal activity was observed (Ravassard et al., 2013).
Thus place field representations likely rely on a combination of entorhinal inputs to hippocampus that include grid, head direction and border cells. Grid cells, like place cells, fire at specific locations of the environment, but they have multiple firing fields that form a triangular grid, and are located in clusters in medial entorhinal cortex and in pre‐ and para‐subiculum (Boccara et al., 2010; Fyhn, Hafting, Treves, Moser, & Moser, 2007; Hafting et al., 2005; Heys, Rangarajan, & Dombeck, 2014).
Each head direction cell fires preferentially when the animal is facing a certain direction, thus head direction cells represent allocentric heading independent of location.
They are found in the dorsal pre‐subiculum and entorhinal cortex, but also in other areas including the anterior dorsal thalamic nucleus and the RSC (Taube, 2007; Taube et al., 1990).
While the head direction system seems to be essential for place cell stability (Calton et al., 2003; Harland et al., 2017), selective elimination of the grid input while retaining hippocampal place fields appears to be possible (Koenig, Linder, Leutgeb, & Leutgeb, 2011). However, how these diverse input patterns are integrated in hippocampal circuits to give rise to spatial and non‐spatial codes remains elusive (Danielson et al., 2016; Lovett‐Barron et al., 2012; Poucet et al., 2013).
Hippocampal theta oscillation (4–12 Hz) is generated by a subcortical network in which the MS likely plays a crucial pacemaker role (Fuhrmann et al., 2015; Hangya, Borhegyi, Szilagyi, Freund, & Varga, 2009; Mamad, McNamara, Reilly, & Tsanov, 2015).
It has been linked to cognitive and other processes (Korotkova et al., 2018) among which exploration (Gangadharan et al., 2016), moving and running, with a direct correlation to speed, (Bender et al., 2015; Sheremet, Burke, & Maurer, 2016) and memory consolidation (Boyce, Glasgow, Williams, & Adamantidis, 2016) are important to spatial navigation. Indeed, septally induced theta rhythm is thought to carry linear velocity information, since this structure has speed cells, its activation/deactivation can initiate/stop locomotion (Fuhrmann et al., 2015), and paces theta in correlation with speed (Tsanov, 2017).
Combining the cognitive map with movement information may allow updating the estimate of self‐position while moving; indeed, speed modulation of hippocampal theta frequency correlates with spatial memory performance in a spatial alternation task (Richard et al., 2013).
This process may be mediated both by place and grid cells that could integrate spatial signals and velocity‐dependent theta waves both in rodents (Chen, King, Burgess, & O’Keefe, 2013) and humans (Bohbot, Copara, Gotman, & Ekstrom, 2017). Moreover, septal inactivation abolishes theta and disrupts grid cell coding (Brandon et al., 2011; Koenig et al., 2011), causes deficit in spatial working memory (Ma et al., 2009) and impairs rats’ ability to estimate linear distances based on self‐motion information (Jacob, Gordillo‐Salas, et al., 2017), while hippocampal place fields are maintained (Mizumori, McNaughton, Barnes, & Fox, 1989).
Selective ablation of septal cholinergic neurons reduced the amplitude and spectral power of theta oscillation without eliminating it (Lee, Chrobak, Sik, Wiley, & Buzsáki, 1994; Rastogi, Unni, Sharma, Laxmi, & Kutty, 2014), while cholinergic stimulation enhanced hippocampal theta (Vandecasteele et al., 2014). Cholinergic M1 receptors located on hippocampal pyramidal neurons might be relevant in this process, being critical for hippocampal synaptic plasticity, theta generation and spatial memory performance in a Y‐maze (Gu, Alexander, Dudek, & Yakel, 2017). At the same time, the potential role of local intra‐septal connections has also been emphasized (Dannenberg et al., 2015; Yang et al., 2014; Zant et al., 2016).
The posterior parietal cortex
The PPC occupies the caudal part of the lateral cortex between the primary somatosensory area and the parieto‐occipital sulcus.
It is generally regarded an associational cortical region, combining inputs from sensory cortices of multiple modalities with top‐down prefrontal inputs and buttom‐up subcortical proprioceptive and vestibular signals (Whitlock, 2017).
Posterior parietal cortex is involved in representing bodily position and spatial orientation, which are key features for understanding its role in navigation.
In one of the first relevant studies, subjects familiarised with a complex virtual town had to navigate to an unseen target area while they underwent positron emission tomography (Maguire et al., 1998).
Along with the hippocampus, parietal cortex was found to be activated when subjects had to perform a sequence of turns to reach a target, irrespective of whether they were performing difficult way‐finding tasks or simply following arrows towards the goal.
Thus the parietal cortex may compute the correct body turns to enable navigating along a route, suggesting it has a crucial role in route creation using proximal‐egocentric cues, in contrast to the hippocampus that mainly represents allocentric maps.
Route learning using proximal salient cues was subsequently associated with a network of structures including the PPC, left medial frontal gyrus and left RSC.
Subjects learnt to navigate in a VR maze with several landmarks at the crossroads; posterior inferior parietal regions showed increasing activation across sessions, correlated with behavioural measures of route expertise (Wolbers, Weiller, & Büchel, 2004).
In an elegant study, London taxi drivers were imaged while navigating passengers in a detailed, topographically accurate videogame reproduction of the British capital. Parietal cortex was activated when the cab driver decided to change his route adapting it to environmental contingencies (e.g. change to a faster lane), confirming the role of PPC in ego‐centric route planning (Spiers & Maguire, 2006).
Selective representation of navigationally salient egocentric information was recently found in the precuneus, part of the posterior medial parietal cortex (Chadwick, Jolly, Amos, Hassabis, & Spiers, 2015).
Similar conclusions were drawn from studying the consequences of PPC lesions. Patients with parietal lesions due to infarction had to navigate through a VR park (i.e. an open environment rich in landmarks) and a VR maze (i.e. a series of identical corridors and intersections) to reach a virtual gold pot; they were impaired on the latter, further corroborating that the impacted parietal area might be involved in processing egocentric cues (Weniger, Ruhleder, Wolf, Lange, & Irle, 2009).
In another study, patients could recall a detailed image of their home city and make distance and proximity judgments but could not navigate between known locations or determine the correct sequence of landmark positions (Ciaramelli, Rosenbaum, Solcz, Levine, & Moscovitch, 2010).
These results further suggest that the PPC is crucial for accessing remote spatial memories within an egocentric reference frame that enables both navigation and re‐experiencing.
Parallel to the above studies of parietal damage in human patients, rats with PPC lesions were found to be impaired on navigation based on proximal but not distal cues in the MWM (Save & Poucet, 2000).
A partial explanation was found later by the same group, observing that rotation of proximal cues in an open field could not elicit a consequential shift in all hippocampal place fields after parietal lesions (Save, Paz‐Villagran, Alexinsky, & Poucet, 2005). These finding suggest a functional importance of the PPC in processing the local spatial frame of reference.
A finer‐grained insight on the complex role of this area may be gleaned from electrophysiological studies. PPC neurons were recorded in rats navigating in a maze that allowed reaching the goal through different routes (Nitz, 2006).
Neuronal discharge reflected a combination of movement direction, spatial position and behavioural information such as left and right turns, often scaled by path segment size, suggesting a relationship with route progression. PPC neurons were also recorded while rats were traversing a squared spiral track (Nitz, 2012).
Firing activity could simultaneously encode three frames of reference (segments, loops and routes), and could thus discriminate analogous segment positions in different loops while still representing reoccurring patterns across segments and loops; therefore, the PPC may help process reference frames for route computing.
Another experiment confirmed that PPC is tuned both to allocentric and egocentric reference frames. Some neurons encoded the egocentric position of a light cue toward which the rat had to move, while others represented this conjunctively with the animal’s head direction, consistent with the role of this area in orienting the body for goal‐directed route progression (Wilber, Clark, Forster, Tatsuno, & McNaughton, 2014). Single‐ and multi‐unit activity revealed that PPC is also tuned to self‐motion (Whitlock, Pfuhl, Dagslott, Moser, & Moser, 2012; Wilber, Skelin, Wu, & McNaughton, 2017).
Activity patterns were found to reoccur in post‐experience sleep in a compacted fashion, temporally coordinated with hippocampal reactivation, suggesting a role in memory consolidation during sleep (Wilber et al., 2017). The role of PPC neurons in encoding position and heading has also been shown recently in head‐fixed mice performing a two‐alternative forced‐choice visual detection task by walking in a virtual T Maze (Krumin, Harris, & Carandini, 2017).
final choice could be predicted from the heading angle with increasing accuracy as the mouse reached the end of the main corridor. PPC neurons appeared to be selective for specific combinations of the animal’s position and of its heading angle (“position‐heading field”).
The retrosplenial cortex
The RSC is a transition area in the posterior cingulate region that links limbic memory areas such as the hippocampus and cortical areas relevant to spatial navigation and behavioural processing of the dorsal stream coordinating visual and motor information (Miller, Vedder, Law, & Smith, 2014).
RSC and PPC are densely interconnected, and are thought to cooperate in coordinating egocentric and allocentric information processing (Clark, Simmons, Berkowitz, & Wilber, 2018).
Patients with damages to the area, typically as a consequence of cerebral haemorrhages or tumour in the splenium of the corpus callosum, consistently show difficulty acquiring new information and retrieving recent autobiographical memories (Gainotti, Almonti, Di Betta, & Silveri, 1998; Osawa, Maeshima, Kubo, & Itakura, 2006; Valenstein et al., 1987).
This has recently been confirmed in primates by within‐subject comparison of memory performance before and after controlled retrosplenial cortical lesions (Buckley & Mitchell, 2016).
Damage involving the RSC can also cause a selective topographical disorientation: in most of these cases, patients recognize familiar landmarks or visual scenes but fail to describe routes between locations, draw a path or navigate efficiently even in familiar environments or learn to navigate in novel settings, indicating that they are unable to derive directional information from landmark cues (Greene, Donders, & Thoits, 2006; Ino et al., 2007; Maeshima et al., 2001; Maguire, 2001; Takahashi, Kawamura, Shiota, Kasahata, & Hirayama, 1997). Notably, patient TT, a former London taxi driver with bilateral hippocampal damage, was impaired at navigating in familiar and novel environments but, unlike patients with RSC lesions, could maintain a sense of direction and his ability to orient in familiar environments (Maguire et al., 2006).
Studies using fMRI support the prominent role of RSC in processing the spatial relationships of landmarks, especially in connection with orientation in a novel environment (Auger, Mullally, & Maguire, 2012; Auger, Zeidman, & Maguire, 2015, 2017; Dilks, Julian, Kubilius, Spelke, & Kanwisher, 2011). Moreover, RSC is involved in processing heading direction (Marchette, Vass, Ryan, & Epstein, 2014; Shine, Valdés‐Herrera, Hegarty, & Wolbers, 2016) and spatial position on the basis of self‐motion cues or path integration (Sherrill et al., 2013, 2015; Wolbers & Büchel, 2005).
Thus, it appears that the RSC supports allocentric representations by processing the stable features of the environment and their spatial relationships, but also enables us to localise ourselves in the environment. Along these lines, in a seminal review article Byrne et al. suggested that RSC is a “translational” area, transforming allocentric representations into egocentric representations and vice versa, allowing the formation of a comprehensive and complete spatial map (Byrne, Becker, & Burgess, 2007).
In rodents, lesions to the RSC impaired allocentric spatial memory in the MWM (Czajkowski et al., 2014; Sutherland, Whishaw, & Kolb, 1988; Vann & Aggleton, 2002), while overexpressing CREB in RSC resulted in spatial memory enhancements (Czajkowski et al., 2014). In addition, tests relying on path integration, integration of egocentric and allocentric information or switching between the two were also affected (Elduayen & Save, 2014; Nelson, Hindley, Pearce, Vann, & Aggleton, 2015; Nelson, Powell, Holmes, Vann, & Aggleton, 2015; Pothuizen, Davies, Aggleton, & Vann, 2010).
Combining a head‐fixed locomotion assay with Ca2+‐imaging in mouse RSC, a population of neurons located predominantly in superficial layers showed activity resembling that of hippocampal CA1 place cells during the same task, while they fired in sequences during movement showing firing fields that form a sparse, orthogonal code correlated with spatial context (Mao, Kandler, McNaughton, & Bonin, 2017).
In another study, rats were trained on a T‐maze task in which the reward location was explicitly cued by a flashing light and RSC neurons were recorded as the rats learned. Most RSC neurons rapidly encoded the light cue, and this representation was not sensitive to the location of the light. However, some neurons encoded also the reward and its location, and they showed distinct firing patterns along the left and right trajectories to the goal (Vedder, Miller, Harrison, & Smith, 2016).
Head‐direction cells (HD), while present in other cerebral areas such as the anterior thalamus, striatum, entorhinal cortex and subiculum, represent about 8% of the RSC population, equally distributed across the granular and dysgranular layers. Some of these cells’ activity is modulated by the velocity of locomotion, while others are tuned to particular combinations of location, direction and movement (Chen, Lin, Green, Barnes, & McNaughton, 1994; Cho & Sharp, 2001).
Those cells are reciprocally connected with the antero‐dorsal thalamus and influence HD firing to preferred direction (Clark, Bassett, Wang, & Taube, 2010). Interestingly, when rats were exploring two connected compartments containing landmarks in reversed orientation, causing conflict between global and local directional cues, some neurons in the dysgranular layer fired facing one direction in one chamber and the opposite in the other.
This indicates that local environmental cues could prevail over head direction information in some neurons, eventually allowing association or dissociation of landmark cues from the head direction signal (Jacob, Casali, et al., 2017).
Recording from rats also confirmed the relevance of RSC in path integration and the integration of allocentric and egocentric elements. Neurons were recorded while the animal was traversing a route that required a specific sequence of body turns depending on the direction in which it was running in a room rich in distal environmental cues.
The animals were making specific turn sequences (route‐based frames) while exposed to distal visual cues (allocentric frame).
A population of neurons was found to code the animal’s allocentric position in conjunction with the progress through the current route as well as left vs right turning behaviour (Alexander & Nitz, 2015).
More recently, the same authors also found populations of RSC neurons that encoded route‐segments as well as the relative position of these segments within an allocentric framework (Alexander & Nitz, 2017).
Rats were trained to navigate a track with a recursive structure; some neurons exhibited periodic activation patterns repeated across similarly shaped route segments, while a larger population exhibited periodicity over the full route, defining a framework for encoding sub‐route positions relative to the whole. This hints at the involvement of RSC in the extraction of path components and coding their spatial relationships.
More information: A sense of space in postrhinal cortex, Science 12 Jul 2019: Vol. 365, Issue 6449, eaax4192 , DOI: 10.1126/science.aax4192 , https://science.sciencemag.org/content/365/6449/eaax4192
Journal information: Science
Provided by Dartmouth College
