Get 20M+ Full-Text Papers For Less Than $1.50/day. Subscribe now for You or Your Team.

Learn More →

Retrosplenial cortex and its role in spatial cognition:

Retrosplenial cortex and its role in spatial cognition: Retrosplenial cortex is a region within the posterior neocortical system, heavily interconnected with an array of brain networks, both cortical and subcortical, that is, engaged by a myriad of cognitive tasks. Although there is no consensus as to its precise function, evidence from both human and animal studies clearly points to a role in spatial cognition. However, the spatial processing impairments that follow retrosplenial cortex damage are not straightforward to characterise, leading to difficulties in defining the exact nature of its role. In this article, we review this literature and classify the types of ideas that have been put forward into three broad, somewhat overlapping classes: (1) learning of landmark location, stability and permanence; (2) integration between spatial reference frames; and (3) consolidation and retrieval of spatial knowledge (schemas). We evaluate these models and suggest ways to test them, before briefly discussing whether the spatial function may be a subset of a more general function in episodic memory. Keywords Learning, memory, cingulate cortex, primate, hippocampal formation, thalamus, neuroimaging, default mode network, immediate-early genes, electrophysiology Received: 17 September 2017; accepted: 18 December 2017 Introduction Retrosplenial cortex (RSC) has fallen within the scope of mem- classes: first, it is involved in the setting of perceived landmarks ory research for at least 40 years (Vogt, 1976) and yet as Vann into a spatial reference frame for use in orientation (spatial and et al. (2009) pointed out in their recent comprehensive review, directional) as well as evaluation of landmark stability; second, it little was discovered about the structure for the first 90 years after stores and reactivates associations between different processing Brodmann first identified it. Since the early 1990s, a growing modes or reference frames for spatial navigation; and third, it has body of evidence has implicated the RSC variously in spatial a time-limited role in the storage and possibly retrieval of hip- memory, navigation, landmark processing and the sense of direc- pocampal-dependent spatial/episodic memories. We conclude tion, visuospatial imagery and past/future thinking, and episodic with some suggestions about how to further refine, and perhaps memory. Early results were difficult to interpret in the absence of ultimately synthesise, these models. precise neuroanatomical, behavioural, electrophysiological and functional data. However, as a consequence of intense research on the RSC, both across animal models using a variety of meth- ods and also in human neuropsychological and imaging studies, Department of Experimental Psychology, University of Oxford, Oxford, a group of theories is now emerging that highlight the involve- UK ment of the RSC in aspects of cognition that go beyond, yet at the Department of Molecular and Cellular Neurobiology, Nencki Institute of Experimental Biology, Warsaw, Poland same time still underlie, our abilities to process spatial informa- Institute of Behavioural Neuroscience, Division of Psychology and tion and retrieve memories. This review will examine the experi- Language Sciences, University College London, London, UK mental data in light of its contribution to spatial cognition, School of Psychology, Cardiff University, Cardiff, UK beginning with a review of the anatomy and connectivity, fol- lowed by functional investigations based on lesion studies, imag- Corresponding author: ing and electrophysiology, and concluding with evaluation and Anna S. Mitchell, Department of Experimental Psychology, University classification of the main ideas that have emerged. We suggest of Oxford, Tinsley Building, Mansfield Road, Oxford OX1 3SR, UK. that the proposals about RSC function fall into at least three Email: anna.mitchell@psy.ox.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 Figure 1. Schematic of the RSC as seen in midsagittal section and located just posterior to the corpus callosum, in humans, rhesus monkeys and rats. Source: Figure by Jeffery (2017); available at: https://doi.org/10.6084/m9.figshare.5414179.v1 under a CC-BY 4.0 licence. the posterior cingulate cortex in primates (Kobayashi and Amaral, 2003). Cortical connections As shown in Figure 2, neural connections of the RSC from the cortex include the parahippocampal region (postrhinal cortex in rodents) (Suzuki and Amaral, 1994), medial entorhinal cortex (Czajkowski et al., 2013; Insausti and Amaral, 2008; Insausti et al., 1987; Jones and Witter, 2007; Van Hoesen and Pandya, 1975) and cingulate cortex (Jones et al., 2005). RSC receives unidirectional inputs from the CA1 field of the hippocampus (Cenquizca and Swanson, 2007; Miyashita and Rockland, 2007) and from the subiculum (Honda and Ishizuka, 2015; Wyss and Van Groen, 1992). It is also interconnected with the extended hippocampal complex, including the presubiculum, postsubicu- lum and parasubiculum (Kononenko and Witter, 2012; Wyss and Figure 2. A schematic diagram detailing the gross connectivity Van Groen, 1992), visuospatial cortical association areas (mainly of retrosplenial cortex. As depicted in the figure, RSC serves as an medial precuneate gyrus, V4 of the occipital lobes and the dorsal interconnected hub for neocortical, hippocampal, parahippocampal bank of the superior temporal sulcus) (Passarelli et al., 2017) and thalamic regions that are functionally involved in the processing and prefrontal cortex (with the heaviest terminations in the dor- of mammalian perceptions important for direction, location, landmarks solateral prefrontal cortex, frontopolar area 10 and area 11 of the and navigation. Different shading is used for effect only. orbitofrontal cortex); these frontal connections are all recipro- cal. RSC also receives inputs directly from V2 of the occipital lobes. There are also prominent excitatory reciprocal connec- Anatomy and connectivity of RSC tions between RSC and posterior secondary motor cortex – namely M2, that have been recently identified in mice (Yamawaki In human and non-human primates, RSC conforms to the cortical et al., 2016). regions that Brodmann identified as areas 29 and 30, which – along with areas 23 and 31 – form part of the posterior cingulate cortex, lying immediately posterior to the corpus callosum Subcortical connections (Figure 1 – left and middle). Rodents lack areas 23 and 31, and RSC itself is located more dorsally and reaches the brain surface In addition, as shown in Figure 2, RSC has major reciprocal (Figure 1 – right). Its central location makes it pivotally posi- subcortical interactions with the anterior (ATN) and laterodor- tioned to receive information from, and readily influence, many sal thalamic nuclei (Aggleton et al., 2014; Kobayashi and key brain regions responsible for the processing of spatial Amaral, 2003, 2007; Van Groen and Wyss, 1990, 1992a and information. 1992b, 2003; Vogt et al., 1987). While the RSC projections to Typically, structural neural connections have been mainly thalamus mainly arise from layer 6, projections from areas 29 derived from studies in animal models (rodents and non-human and 30 provide different densities of terminal fields in the three primates), while the majority of neural connections studied in subdivisions – anteroventral, anteromedial and anterodorsal – humans have been derived functionally. It is known that in of the ATN (Aggleton et al., 2014). Given that the ATN and both rats and primates, the majority of RSC (RSC granular A laterodorsal thalamic nuclei provide major RSC inputs, it is of and granular B, and RSC dysgranular) connections (up to 78%) interest to establish where these two thalamic structures receive originate in or are received from other parts of RSC and from their inputs. Briefly, the laterodorsal thalamus receives inputs Mitchell et al. 3 from the postsubiculum, visual association cortex and the lat- deficits and found that involvement of the RSC was prominent in eral mamillary bodies (Shinkai et al., 2005; Sripanidkulchai and impairments of landmark processing, particularly when it came Wyss, 1986; Taube, 2007; Vogt and Miller, 1983), while the to reporting distances and directions between known landmarks ATN receives inputs directly from the lateral and medial mamil- or describing the positions of known landmarks or buildings on a lary bodies and from the hippocampal formation/ subicular map. complex. Possibly, the key message transmitted from the lateral Experimental lesion studies in non-human primates can be mamillary bodies to the anterodorsal subdivision of the ATN much more precise, and also bilateral, which has provided new and the laterodorsal thalamus is information about the position insights into RSC function. In rhesus macaque monkeys, damage of the head received from the dorsal tegmental nucleus of to the RSC, which included the most caudal part of the posterior Gudden located in the midbrain (Cajal, 1955; Guillery, 1956, cingulate cortex, selectively impaired the ability to retrieve 1957; Powell et al., 1957; Taube, 1995, 2007). In contrast, we object-in-place scene discriminations that the monkeys had pre- do not yet fully know what information is transmitted to the viously learnt (retrograde memory) (Buckley ad Mitchell, 2016). RSC and cingulate cortex via the anteromedial and anteroven- In these tasks, animals have to learn and remember the location tral subdivisions of the ATN, although theta-modulation (Vertes of a discrete object in a spatial scene. In contrast, these same et al., 2001) and theta-modulated head direction (HD)-signalling animals were able to learn new object-in-place scene discrimina- neurons have been identified in the anteroventral subdivision in tions postoperatively (anterograde memory), so their ability to rats (Tsanov et al., 2011), and gravity-tuned neurons have been organise spatial information appeared to remain intact. However, identified in primate ATN (Laurens et al., 2016). In addition to during new learning that involved a 24-h delay period between the above major connections, there are also lesser connections successive sessions of learning the new set of object-in-place dis- with the mediodorsal thalamus and rodent lateral posterior tha- criminations (i.e. from session 1 to session 2), monkeys with lamic nucleus (Aggleton et al., 2014; Powell, 1978). RSC also RSC damage made more errors than controls during postopera- receives inputs from the intralaminar thalamic nuclei (impor- tive session 2 of new learning only. This selective deficit, which tant for arousal) and primate medial pulvinar (supporting visual was present in all monkeys with RSC damage, comprised a spe- attention) (Baleydier and Mauguiere, 1985; Buckwalter et al., cific impairment in their ability to retrieve these new discrimina- 2008; Vogt et al., 2006). tions which they had seen only 24 h beforehand. The task, In general, the anatomy shows that RSC interacts reciprocally object-in-place scene discriminations, incorporates elements of with many brain regions, consistent with its role, described both spatial (e.g. landmark information) and episodic-like mem- below, in a number of core cognitive competences. In particular, ory (unique object-in-place scene discriminations, with one of it is clear that the RSC interacts with many visual areas of the the objects in each discrimination paired with a reward if it is brain across mammalian species. Of interest is the more unidirec- selected) without being explicitly autobiographical in nature tional relationship with hippocampus and with perirhinal cortex. (Gaffan, 1994; Mitchell et al., 2008; Mitchell and Gaffan, 2008; Murray and Wise, 2010). The novel findings observed in the monkeys’ performance led the authors to conclude that an intact Lesion studies RSC is particularly important for the ability to retrieve informa- tion that has been previously acquired, regardless of whether The literature on pure RSC lesions in humans is sparse and these memories are autobiographical, or episodic (in the pure mostly from unilateral pathology due to the rarity with which sense of what/ where and when), or actively spatial in nature localised infarcts or injury occur to this region, and so most of (Buckley and Mitchell, 2016). Finally, the ability to retrieve this our knowledge of human RSC comes from neuroimaging, which information did not require the monkeys to move around in their we discuss later. Most of the identified lesion-induced deficits environment, although the successful executions of self-gener- appear to involve memory and spatial processing. Maguire ated hand-eye coordinated movements (in order to select the cor- (2001) conducted a comprehensive review of the literature on rect object within the scene on the touchscreen) were necessary. RSC extant at the time and concluded that case studies of RSC Studies involving smaller mammals have proved vital in fur- lesions reveal deficits in episodic memory (memory for life thering our understanding of the contribution of the RSC to cog- events), occurring particularly following left-sided lesions, but nition, as they afford far greater neuroanatomical precision than also consistent reports of topographical disorientation (getting is currently possible in primate studies. An early study by Berger lost), with or without concomitant memory deficits, most of et al. (1986) found that rabbits with RSC lesions could acquire a which followed right-sided lesions. The area that was most con- tone-light discrimination, but were profoundly impaired in sistently involved in the pure disorientation cases was Brodmann reversing it, suggesting a failure to modify a recently established area BA30. Maguire (2001) noted: ‘In every case, the patient was memory. Given the dense interconnections between the RSC and able to recognise the landmarks in their neighbourhoods and the hippocampal spatial system, the majority of subsequent retained a sense of familiarity …’. Despite this, none of the lesion studies have focused on spatial learning. patients were able to find their way in familiar environments, and Some of the early studies into the effects of RSC lesions on all but one were unable to learn new routes. Studies since then spatial tasks produced mixed results. This divergence in findings have confirmed the link between RSC lesions and topographic may be attributable to methodological considerations such as the disorientation, with association of left-sided infarct with memory use of electrolytic or ablation lesions, which destroy fibres of deficits (Kim et al., 2007) and of right-sided lesions with spatial passage and consequently may exaggerate the impact of the RSC impairment (Hashimoto et al., 2010, 2016), although spatial damage, while other studies spared the more caudal aspect of the impairment has also been reported in patients with left-sided RSC, which is now known to be critically involved in spatial lesions (Ino et al., 2007; Ruggiero et al., 2014). Claessen and van memory (Vann and Aggleton, 2002, 2004). Despite these earlier der Ham (2017) conducted a review of lesion-related navigation 4 Brain and Neuroscience Advances controversies, there is now very good evidence that RSC lesions properties of the test environment or the juxtaposition of highly in rodents disrupt spatial memory. Deficits are consistently salient visual cues. Rats learnt the location of the platform either reported on tasks that involve allocentric spatial processing, par- by actively swimming to the platform or passively, by being ticularly when – as with the imaging studies – visual cues are repeatedly placed on the platform location. They were then given needed for orientation (Hindley et al., 2014). Such tasks include a test in which they had to swim to the correct location for the learning the fixed or alternating location of a platform in the first time. RSC-lesioned rats were selectively impaired in the Morris watermaze (Sutherland et al., 1988; Vann and Aggleton, passive condition, indicating that RSC damage did not disrupt 2002, 2004; Whishaw et al., 2001), the radial arm maze (Keene navigation per se, but selectively impaired the ability to switch and Bucci, 2009; Pothuizen et al., 2008; Vann and Aggleton, spatial frames of reference and different spatial viewpoints when 2004) and object-in-place discriminations (Parron and Save, navigating to the platform from a novel position in the environ- 2004). There is some evidence that the RSC dysgranular region ment (Nelson et al., 2015a). Similarly, complete RSC or selective (area 30; see Figure 1 – left) may be particularly important for RSC dysgranular lesions disrupted the ability to recognise the processing allocentric space, as rats with selective RSC dysgran- layout of a room from different viewpoints (Hindley et al., 2014). ular lesions were unable to use distal visual cues to guide spatial Taken together, RSC effects appear to depend on the extent to working memory and relied instead on motor sequence informa- which task performance relies on the retrieval of spatial land- tion (Vann and Aggleton, 2005). Furthermore, deficits have also marks for orientation, or the need to switch between different been found on tasks that require the use of directional informa- spatial strategies or viewpoints. This is in line with the proposal tion (Keene and Bucci, 2009; Pothuizen et al., 2008; Vann and that key aspects of RSC functioning include integration of the Aggleton, 2004) as well as self-motion cues (Elduayen and Save, context in which an event occurs, learning about the significance 2014; Whishaw et al., 2001). In some instances, the involvement of such stimuli or updating representations as new information of RSC has been found to be time-limited: for example, Maviel comes on-line. et al. (2004) found that RSC inactivation in mice disrupted the retrieval of a recent 1-day-old spatial memory but not a remote Brain imaging (positron emission 30-day-old one, while Keene and Bucci (2009) found large impairments on radial maze performance for a 30-s delay relative tomography, functional magnetic to a 5-s delay. Findings such as these, combined with the imme- resonance imaging and immediate- diate-early gene study findings described later, and the primate early gene activation) studies mentioned above, suggest a particular role for RSC when spatial information needs to be retrieved from memory. As outlined above, human, primate and rodent RSC lesion stud- In general, the magnitude of spatial deficits after RSC lesions ies have pointed to a role in spatial processing: complementary tends to be smaller and less striking than the spatial impairments evidence comes from research using metabolic brain imaging, associated with either hippocampal or ATN damage. The most particularly positron emission tomography (PET), functional striking demonstration of this difference is T-maze alternation magnetic resonance imaging (fMRI) and immediate-early gene performance, which is acutely sensitive to both hippocampal and activation (IEG) studies. ATN damage (Aggleton et al., 1986, 1996), but is often spared Human neuroimaging studies have been complicated by the after RSC lesions (Meunier and Destrade, 1988; Neave et al., lack of agreement about exactly which regions belong to RSC 1994; Nelson et al., 2015b; Pothuizen et al., 2008). Indeed, the proper. While the scene-selective posterior and ventral bank of full impact of RSC lesions often only emerges under specific the parieto-occipital sulcus is often referred to as RSC, Silson conditions or when animals are required to shift between differ- et al. (2016) have suggested that the term be reserved for the ent spatial metrics. For example, temporary inactivation of the region within the callosal sulcus extending onto the isthmus of RSC selectively impairs navigation in the dark, but not the light the cingulate gyrus. Such distinctions are relevant for the issue of (Cooper et al., 2001). However, Wesierska et al. (2009) found the specificity of RSC processing, as well as its cross-species that rats with RSC dysgranular lesions could learn to avoid the homology, which is still not fully established. shock zone of a rotating platform if the rotation occurred in the In an early PET study of cerebral glucose metabolism, dark, so darkness per se does not seem to be the problem. The rats Minoshima et al. (1997) found reduced activation in the posterior could also learn to avoid the shock zone if this was defined by cingulate in patients with mild cognitive impairment and early allocentric room cues provided there were no conflicting local Alzheimer’s disease, while Nestor et al. (2003) found that the cues; thus, there was not a straightforward impairment of allo- RSC part of the posterior cingulate, was the most consistently centric cue use either. There was a notable impairment when the hypometabolic region. More recent imaging studies have contin- animals had to disregard the local cues and focus on the room ued to confirm that changes in glucose metabolism in the poste- cues. Thus, as the authors noted, impairments arose when rele- rior cingulate cortex, as well as hippocampal atrophy, are early vant and irrelevant cues needed to be segregated. Similarly, biomarkers for Alzheimer’s disease and are likely present many impairments on both the radial arm maze and T-maze often only years before the clinical symptoms appear (e.g. An et al., 2017; emerge when intra-maze cues are placed in conflict with extra- Teipel et al., 2016). maze cues (Nelson et al., 2015b; Pothuizen et al., 2008; Vann and Since the advent of fMRI in cognitive neuroscience, many Aggleton, 2004). studies have investigated RSC activation as subjects perform A further illustration of the selective nature of RSC lesion- tasks in the scanner. Indeed, RSC is now considered to be part of induced spatial deficits comes from an experiment by Nelson the so-called default mode network, which consists of a set of et al., 2015a in which the location of a submerged platform in a brain structures including medial frontal and medial temporal Morris watermaze was determined by either the geometric lobe regions, lateral and medial parietal areas and the RSC (Vann Mitchell et al. 5 et al., 2009), which are active when subjects are not performing when subjects were shown stationary views of the environment a task in the scanner but rather are lying in the scanner at ‘rest’, and had to make orientation judgements (Shine et al., 2016). or actively simulating a situation (particularly one close in time Furthermore, recent work has also examined RSC activation and space to the present (Tamir and Mitchell, 2011), or when they when participants navigate in a virtual 3D environment (Kim are retrieving a memory (Sestieri et al., 2011)). et al., 2017). Interestingly, in this study, the RSC activation was Cognitive tasks that reliably activate RSC in fMRI studies particularly sensitive to the vertical axis of space, which the include most that have a spatial component, especially when this authors suggest may be supporting processing of gravity, which requires use of the visual environment to retrieve previously is a directional cue in the vertical plane and may be useful for 3D learned information in order to orient. These typically involve navigation. Given that there is evidence for both local and global virtual reality simulations in which subjects navigate, by joystick encoding of direction in RSC, the question arises as to how these or sometimes just by imagination, around a virtual environment, might both be accommodated within the one structure; we return such as a town. In one of the earliest studies, Wolbers and Büchel to this question later. (2005) scanned subjects as they learned a virtual maze-like town While the foregoing studies looked at global spatial environ- and found that RSC activation increased steadily with learning ments, work from the Maguire lab has suggested a role for RSC and paralleled increasing map performance. Similarly, in a study in the processing of individual landmarks. Auger et al. (2012) of London taxi drivers in a virtual environment based on real scanned subjects as they viewed a variety of images with a mix- maps of London (Spiers and Maguire, 2006), RSC activation ture of large and smaller objects and found that RSC was acti- occurred during route planning, spontaneous trajectory changes vated only by the spatially fixed, landmark-like objects, and and confirmation of expectations about the upcoming features of furthermore that the extent of activation correlated with naviga- the outside environment - but not, interestingly, expectation vio- tion ability. In a follow-up study using MVPA, Auger and lations. Another fMRI study confirmed that RSC activity was Maguire (2013) showed that decoding of the number of perma- specifically associated with thoughts of location and orientation, nent landmarks in view was possible, and more so in better navi- as opposed to context familiarity or simple object recognition gators, concluding that RSC, in particular, is concerned with (Epstein et al., 2007). In both studies, the overall pattern of RSC encoding every permanent landmark that is in view. They then activation differed from the one observed for hippocampus (Iaria showed that this RSC permanence encoding also occurred when et al., 2007), with the entire RSC active during both encoding and subjects learned about artificial, abstract landmarks in a feature- retrieval of spatial information. less Fog World (Auger et al., 2015), demonstrating that the RSC A related line of work has investigated the encoding of loca- is involved in new learning of landmarks and their spatial stabil- tion and/or direction by RSC. Marchette et al. (2014) performed ity and also that such learning correlates with navigation ability multi-voxel pattern analysis (MVPA) of fMRI brain activation (Auger et al., 2017). Puzzlingly, however, the involvement of patterns on subjects recalling spatial views from a recently RSC seems better correlated with the stability per se than with learned virtual environment. Because MVPA compares fine- the orientational relevance of the landmarks (Auger and Maguire, grained patterns of activation, it allows inferences to be made 2018a). about whether a subject is discriminating stimuli. The virtual Some meta-analyses of human imaging studies have indicated environment comprised a set of four museums located near each that higher RSC activation occurs when subjects process land- other in a virtual park. RSC activity patterns were similar when mark information (Auger et al., 2012; Auger and Maguire, 2013; subjects faced in similar directions and/or occupied similar loca- Maguire, 2001; Mullally et al., 2012; Spiers and Maguire, 2006) tions within each museum, suggesting that RSC was activating and associate the current panoramic visual scene with memory the same representations of local place and local direction, even (Robertson et al., 2016). Further evidence has revealed that RSC though the environments were separated and oriented differently is activated when subjects retrieve autobiographical memories in global space. Similarity judgement reaction times were faster (Maddock, 1999; Spreng et al., 2009) or engage in future think- for homologous directions or locations, suggesting encoding by ing or imagining (Tamir and Mitchell, 2011), although RSC local features independent of global relationships. However, it appears more engaged with past than future spatial/contextual was not demonstrated that subjects had been able to form global thinking (Gilmore et al., 2017). While the retrieval of autobio- maps of the virtual space (i.e. the reference frame in which the graphical memories, imagining and future thinking may not local spaces were set), so the question remains unanswered about explicitly engage spatial processes, they are nonetheless closely whether RSC is also involved in relating directions within a allied to the spatial functions of RSC and its identified role in global space. retrieval, as they require self-referencing to spatial contexts and Robertson et al. (2016) also found encoding of local land- the updating of spatial representations as events are recalled or marks in a setting in which subjects viewed segments of a 360° imagined based on subjective memories. panorama that either did or did not overlap. RSC activation was Animal models, in particular rodent experiments that engage higher when subjects subsequently viewed isolated scenes from their ability to readily explore their spatial environment, have the overlap condition and judged whether it came from the left or provided imaging evidence across mammalian species that high- the right side of the panorama. A study by Shine et al. (2016) did, lights the importance of the RSC for spatial functioning. One however, find evidence for global heading representation in RSC. particular experimental approach is to study RSC functioning They investigated RSC and thalamus activation in subjects who in the intact rodent brain by investigating the extent, and loca- had learned a virtual environment by walking around with a tion, of the activation of learning-induced immediate-early head-mounted display, which provides vestibular and motor cues genes (IEGs; e.g. Arc, Fos or Zif268) after animals have per- to orientation. They found activation of both structures, which formed a behavioural task. Most of these studies have shown both contain directionally sensitive HD cells (discussed below), increased expression of IEGs in the RSC as a consequence of 6 Brain and Neuroscience Advances spatial learning (Maviel et al., 2004; Vann et al., 2000). One of of landmarks within the room as a whole. Some cells behaved the distinct advantages of this approach is that it allows for far like typical HD cells and fired whenever the animal faced in a greater anatomical precision, for example, revealing subregional particular direction in the global space, while others fired in one or layer-specific differences in RSC activity after animals have direction in one compartment and the opposite direction in the performed a spatial task (Pothuizen et al., 2009). other compartment, as if these cells were more interested in local IEG studies have also revealed the involvement of RSC in direction than global direction. This observation is thus reminis- spatial memory formation. Tse et al. (2011) investigated the two cent of the fMRI experiment by Marchette et al. (2014) discussed IEGs, zif268 and Arc, as rats learned flavour-place pairs; they earlier, in which human subjects showed similar RSC activation found up-regulation of these genes in RSC when animals added patterns in local subspaces independent of their global orienta- two new pairs to the set. A more recent approach has been to tion. Together, these results support the idea that RSC might be combine IEG mapping with chronic in vivo two-photon imaging involved in relating spatial reference frames, with some cells to study the dynamics of Fos fluorescent reporter (FosGFP) in responsible for local orientation and others responsible for the RSC dysgranular cortex during acquisition of the watermaze task bigger picture. (Czajkowski et al., 2014). Higher reporter activity was observed More broadly, the findings concerning HD cells suggest that when animals relied on a set of distal visual cues (allocentric RSC neurons may be integrating landmark information coming strategy), as compared to a simple swimming task with one local from the visual cortex, together with the ongoing HD signal being landmark. Moreover, these observations also revealed a small assembled and maintained by more central in the HD network. population of neurons that were persistently reactivated during Such interaction might depend on the strength and/or reliability of subsequent sessions of the allocentric task. This study showed the sensory input (i.e. landmarks) to RSC and/or the HD system that plasticity occurs within RSC during spatial learning and also (Knight et al., 2014), raising the possibility that RSC directional suggested that this structure is critical for formation of the global neurons have the task of evaluating landmarks and deciding representation. Indeed, in another set of experiments, optogenetic whether they are stable and/or reliable enough to help anchor the reactivation of Fos-expressing neuronal ensembles in mouse sense of direction (Jeffery et al., 2016). RSC led to the replication of context-specific behaviours when The above notwithstanding, only around 10% of RSC neurons the animal was in a safe context, devoid of any features of the seem to be HD neurons, the remainder having more complex fir- original training context (Cowansage et al., 2014). ing correlates. Many of these seem related to the actions the ani- Taken together, these complementary human and animal stud- mal is performing. The first systematic analysis by Chen et al. ies highlight that RSC functioning is involved in spatial learning (1994a, 1994b) reported RSC cells related to body turns in addi- and memory, particularly when environmental cues (landmarks) tion to those with spatial firing characteristics. A subsequent are to be used for re-orientation and perhaps navigation. Studies study found RSC cells with firing significantly correlated with of the time course of RSC involvement suggest a dissociation running speed, location and angular head velocity (Cho and between new learning and memory retrieval/updating. The impli- Sharp, 2001). Similarly, cells that respond to specific combina- cation is that perhaps RSC is less involved in spatial perception per tions of location, direction and movement were reported by se, and more involved with visual memory retrieval and editing. Alexander and Nitz (2015), who recorded RSC neurons as rats ran on two identical ‘W’-shaped tracks located at different places in a room. As well as ordinary HD cells, they found cells encod- Single neuron studies ing conjunctions of local position, global position and left/right turning behaviour. In a later study (Alexander and Nitz, 2017), Researchers typically turn to rodent single-neuron studies to some RSC neurons were found to show firing rate peaks that address fine-grained questions about encoding. Chen et al. recurred periodically as animals ran around the edge of a plus (1994a, 1994b) after conducting the first electrophysiological studies of spatial correlates of rodent RSC reported that around maze – some cells activated once per circumnavigation, some 10% of RSC cells in the rat have the properties of HD cells. twice, some four times and so on. Since the environment had These are cells that fire preferentially when the animal faces in a fourfold symmetry, this observation again suggests a possible particular global direction; cells with these properties are found role in relating local and global spatial reference frames. in a variety of brain regions, and are thought to subserve the However, recurring activation patterns having fourfold symmetry sense of direction (Taube, 2007). RSC head direction cells have were also seen when the animal ran on a ring track, with no local very similar properties to those in other regions, although inter- substructure, so it is possible that the cells were responding to estingly they fire slightly in advance of the actual head direction some type of symmetric feature, such as the corners of the room, (Cho and Sharp, 2001; Lozano et al., 2017). However, 90% of the that was present in the distal room cues. RSC neurons had more complex firing correlates, and no clear In contrast to encoding of route, within which every location hypothesis about overall RSC function emerged. that the animal visits along the full trajectory is represented, oth- A later study by Jacob et al. (2017) similarly found a sub- ers have reported encoding of navigational or behaviourally sig- population of HD cells, in the RSC dysgranular cortex only, the nificant cues (e.g. goal-location coding) by RSC in simpler firing of which was controlled by the local environmental cues linear environments. In a study by Smith et al. (2012), animals independently of the global HD signal. They also – like Chen on a plus maze learned to approach the east arm for reward for et al. – found a further sub-population of directionally tuned cells half of each session and then switched to the west. RSC neurons that showed mixed effects, being influenced both by landmarks developed spatially localised activity patches (‘place fields’) and by the global head direction signal. This experiment took that were sensitive to reward-associated locations, and the num- place in an environment composed of two local sub-compart- ber of place fields substantially increased with experience. ments (two connected rectangles) that had opposite arrangements However, unlike co-recorded hippocampal place cells, which Mitchell et al. 7 produce very focal place fields, RSC place fields were dispersed Landmark processing and sometimes covered the entire arms. One function of RSC The first set of views is that RSC has a specific function in the place fields could be enabling the rats to discriminate two behav- encoding of the spatial and directional characteristics, as well as ioural contexts. stability, of landmarks, independent of their identity. This view A recent study by Mao et al. (2017) reported more hippocam- emerges from such findings as that HD-cell sensitivity to land- pal-like activity in RSC cells, finding spatially localised activity marks is reduced following RSC lesions (Clark et al., 2010), that (i.e. place fields) on a treadmill during movements in head-fixed some RSC directionally tuned cells respond to environmental mice. Locations on the track were marked by tactile cues on the landmarks in preference to the main HD network signal (Jacob travelling belt. As with hippocampal place cells, changes in light et al., 2017), that RSC is active when humans process landmark and reward location cause the cells to alter their firing locations permanence (Auger et al., 2012, 2017; Auger and Maguire, 2013) (remap). These observations support the notion that RSC is sensi- and that lesions to RSC in human subjects cause them to lose the tive to spatially informative cues and contextual changes. ability to use landmarks to orient (Iaria et al., 2007). It is also In addition to place-, cue- and reward-location, Vedder et al. supported by findings that rats with RSC lesions are poor at using (2017) reported conjunctive coding. In a light-cued T-maze task, allocentric spatial cues to navigate (Vann and Aggleton, 2005). RSC neurons increased responsiveness to the light cue, mostly By this view, the function of RSC is to process landmarks as cur- irrespective of left–right position, but they also frequently rently perceived and use them to update an already established responded to location or to reward. Responses involved both spatial framework so that in future they can be used for better increased firing (on responding) and decreased firing (off self-localisation and re-orientation. This viewpoint supposes a responding). Interestingly, responding to the light often slightly particular role for landmarks in the ongoing formation and updat- preceded the onset of the light cue, an anticipatory response also ing of spatial representations and is consistent with the close rela- reported earlier in RSC in rabbits in an associative conditioning tionship of RSC to visual areas as well as to the hippocampal task (Smith et al., 2002). The location-sensitive firing on the stem spatial system. Recently, Auger and Maguire suggested that RSC of the T often distinguished forthcoming left and right turns, so- processing of landmarks may be more to do with permanence and called splitter behaviour also seen in hippocampal place cells stability than orientation relevance, and indeed that RSC’s role in (Wood et al., 2000). Thus, although responding is associated with permanence may extend beyond landmarks to other domains a cue, there seems sometimes to be a supra-sensory component (Auger and Maguire 2018a, 2018b). related to a learned expectation. To summarise, then, the results from electrophysiology stud- ies of RSC neurons provide a mixed picture in which spatial pro- Spatial representations cessing dominates, but the nature of the processing is hard to pin down exactly. It is clear that place and heading are represented, The second view, which could be regarded as an extension of the but so are other variables, and the nature and function of the con- first to information beyond landmarks, is that this area serves to junctive encoding remains to be elucidated. mediate between spatial representations, as detailed in the review by Vann et al. (2009). For some investigators, this has meant between egocentric and allocentric processing, although The RSC contribution to spatial egocentric suggests different things to different researchers, meaning self-motion-updated to some and viewpoint-dependent cognition – consensus and to others. Chen et al. (1994a) made the specific proposal that the controversies egocentric information processed by RSC concerns self-motion, The experimental literature reviewed above has revealed areas a position also taken by Alexander and Nitz (2015); this is sup- where investigators are in general agreement, and other areas ported by their and others’ observations that directionally tuned where there is debate or uncertainty. In this section we review neurons in RSC are updated by self-motion (sometimes called these areas and outline some ways forward to resolve these. idiothetic) cues and that navigation in RSC-lesioned animals is There seems to be general agreement that RSC has a role in affected by darkness (Cooper and Mizumori, 1999). Other allocentric spatial processing, as highlighted in the review of authors have taken “egocentric” more broadly to mean spatial Vann et al. (2009), but there are differences in opinion as to the items encoded with respect to the body versus with respect to the exact contribution it makes, and also in whether it has a broader world. For example, Burgess and colleagues have suggested that role in memory, of which space is just a subcomponent. In this RSC is part of the progressive cortical transform of parietal ego- regard, the status of RSC research is a little reminiscent of hip- centrically to hippocampal allocentrically encoded information pocampal research 30 years ago. Our conclusion is that the litera- (Byrne et al., 2007): the hypothesis of egocentric-allocentric ture has yielded three broad, somewhat related views concerning transformation by RSC recurs repeatedly in the literature (see RSC’s spatial function, which we explore further below: Vann et al. (2009) for discussion). Another, not dissimilar view is that RSC is involved in con- 1. It processes landmarks and landmark stability/permanence, structing and relating allocentric spatial reference frames more possibly in service of spatial/directional orientation or per- generally, not necessarily egocentric/allocentric ones. This haps more broadly. view is supported by studies such as the museums study of 2. It mediates between spatial representations, processing Marchette et al. (2014), which found similarities in the encod- modes or reference frames. ing of local spaces even though these were separated in global 3. It is involved in consolidation and retrieval of spatial space, and the similar findings of Jacob et al. (2017) that some schemas, for example to support episodic memory. RSC neurons constructed a directional signal based on local 8 Brain and Neuroscience Advances cues while others used the global space. A related notion is that compares these external cues with internal models of the situa- RSC is involved in switching between different modes (as tion, including input regarding self-motion. opposed to frames) of spatial processing, such as from light to dark (Cooper et al., 2001; Cooper and Mizumori, 1999), or dis- Open questions tal to proximal cues and so on. These models all share the underlying feature that RSC has Resolving the above ideas into a single, inclusive model of RSC access to the same spatial information represented in different function (if this is possible) will require the answering of some of ways, and is needed in order to switch between these. the outstanding questions raised by the studies to date. Below, we outline some of these questions. Spatial schema consolidation/retrieval Does RSC have a specific interest in The final view, which is broadest of all, is that RSC is involved in formation and consolidation of hippocampus-dependent spatial/ landmarks, as a subclass of spatial cue? episodic memories. What differentiates these models from the A finding that has emerged from multiple studies of spatial pro- foregoing, and also from standard theories of hippocampal function, cessing is that RSC is particularly involved in the processing of is the incorporation of a temporal dimension to the encoding. By landmarks, which is to say discrete objects or visual discontinui- this is meant that RSC is not needed for de novo spatial learning, ties in the panorama that serve, by virtue of their distant location but is required when the animal needs to draw on a previously and spatial stability, to orient the sense of direction. However, the learned set of spatial relationships, in order to execute a task or specific hypothesis that it is interested in landmarks as discrete acquire new information to add to its stored representation. objects as opposed to, say, visual panoramas, has not been fully These ideas draw on two sets of theoretical work already tested. An unanswered question then is whether RSC is engaged extant in the literature: the idea proposed by Marr (1971) that rap- during spatial processing in the absence of landmarks; for exam- idly formed hippocampal memories are slowly consolidated in ple, in an environment devoid of discrete spatial cues in which neocortex, and the idea that spatial learning entails the formation geometry or smooth visual shading provides the only cues to not just of task- or item-specific memories, but also of a more direction. It should be noted that most types of geometric envi- general framework within which the memories are situated, which ronment (squares, rectangles, teardrops, etc) have corners, which has sometimes been called a schema (Morris, 2006; Ghosh and could in principle act as discrete landmarks, so care would have Gilboa, 2014). An example of a schema might be the watermaze to be taken with the environmental design to ensure the absence task, in which a rat is faced with needing to learn a new platform of all such discrete visual stimuli. The general question to be location: learning of the new location is faster than the original answered here is whether the brain, via the RSC, treats landmarks because the rat already knows the layout of the room and the as a special category of object or whether the interest of RSC in watermaze, and the procedures required to learn the platform landmarks stems solely from their spatial utility, derived from the location – learning the location for today requires just a small constant spatial relations between them, or from their permanent updating. nature, irrespective of their status as landmarks (Auger and Support for this consolidation/updating idea comes from mul- Maguire, 2018 a and b). tiple observations in the literature that the role of RSC in behav- ioural tasks is frequently time-limited in that its effects occur later in training rather than immediately. In particular, there Does RSC mediate between spatial seems to be a 24-h time window after training, below which RSC representations? is engaged less, but after which it becomes involved: see the The core idea here is that RSC may not be needed for spatial experiment by Buckley and Mitchell (2016), and the observation learning per se, but is needed when the subject moves between by Bontempi and colleagues that IEGs are up-regulated when representational modes. This may entail switching from egocen- mice retrieve a 30-day-old memory, but not a 1-day-old one tric to allocentric encoding of cues, or relating an interior space (Maviel et al., 2004). IEG studies have also revealed greater to an exterior one (e.g. deciding which door one needs to exit engagement for RSC in spaced versus massed training (Nonaka through to reach the carpark). This view is an extension of the et al., 2017), consistent with the need, in spaced training, for local idea discussed above, that RSC is needed to be able to use reactivation of a partly consolidated (10-min-old) memory rather spatial landmarks to retrieve current location and heading. The than a completely newly formed (30-s-old) one. important new ingredient supplied by the reference frame frame- This time-dependency has led several investigators to propose work, as it were, is that at least two representations have had to that RSC is part of a primacy system (Bussey et al., 1996; Gabriel, be activated: for example, being in one place and thinking about 1990), the function of which is to retrieve and process informa- another, or navigating in the dark and remembering where things tion learned earlier. Ranganath and Ritchey (2012) put forward are based on experience in the light. The question to be answered, one of the most detailed of such models that proposed that RSC, therefore, is whether RSC is indeed needed for a subject to acti- together with parahippocampal cortex, form part of a posterior vate two representations simultaneously. cortical network that functions to support episodic memory. They Testing this idea is complicated by the demonstration dis- suggest that this network matches incoming cues about the cur- cussed earlier that RSC is also needed to use local spatial cues to rent behavioural context to what they call situation models, retrieve a previously learned spatial layout. Since it is required which are internally stored representations of the relationships for current self-localisation, the idea that it is also needed for among the entities and the environment. According to their view, spatial imagination, or route planning, or future thinking, or other the parahippocampal cortex identifies contexts and the RSC Mitchell et al. 9 similar imagination-based functions, is hard to test directly will be useful in these studies, as they allow more precise inter- because a lesioned subject can’t even get past the initial orienta- ruption of selective neurons. Given the well-known connectivity tion problem. However, more temporally focused interventions of these structures, several experimental schemes can be pro- such as optogenetic (in)activation may be of help here. For exam- posed. For example, the general population of hippocampal neu- ple, once intact animals have learned a radial maze task, it may be rons sending projections to the RSC could be targeted by both found that RSC is needed if the lights are turned off halfway chemogenetic and optogenetic approaches with a retrogradely through a trial, forcing a switch from one processing mode to transported vector. Since the effective illumination of the entire another. Similarly, rats familiarised with a small space inside a hippocampus would be technically challenging given its subcor- larger one may be able to navigate between the two when RSC is tical location, the chemogenetic approach would be currently operating, but not when it is inactivated. These types of task preferred in this case. The presynaptic terminals of the CA1 pro- probe retrieval and manipulation of already-stored spatial jection neurons could be optically activated within RSC after representations. hippocampal vector injections. In this case, an implantable, light- emitting diode (LED)-based module could be positioned on top of dysgranular RSC to stimulate areas 30 and 29c. With the What is the time course of RSC involvement development of efficient orange LEDs, a similar approach could in spatial learning? be used for inhibition. Red-shifted opsins could further increase the range, potentially allowing the illumination of the entire RSC It remains an open question exactly when RSC comes into play and enabling optogenetic intervention in the hippocampus. during formation and use of spatial memories. IEG studies have Finally, transsynaptic circuit labelling with rabies virus could found that activation occurs rapidly (within minutes) of modify- single-out even more specific sub-populations of projection neu- ing a spatial schema (Tse et al., 2011), and the massed/spaced rons (Callaway and Luo, 2015). In all cases, temporally precise learning experiment from the same group suggests a role for RSC interruption of processing epochs could be achieved. during at least the first few minutes (Nonaka et al., 2017). The extent to which RSC and hippocampus are functionally However other animal studies have found that RSC dependence coupled could also be examined by combining temporary modula- of a task does not appear until memories are reactivated again tion techniques with electrophysiology: for example, assessing the after a delay of up to 24 h (e.g. Buckley and Mitchell, 2016), effect of hippocampal inactivation on neuronal firing within RSC presumably to allow an epoch of time to pass before re-engage- or vice versa. Extant data suggests that RSC inactivation causes ment of the RSC memory can occur. Consequently, as our current changes in hippocampal place fields (Cooper and Mizumori, 2001) understanding lies it is difficult to distinguish if there is a critical and the hippocampal inactivation alters experience-dependent time window in which RSC is engaged by spatial tasks or whether plasticity in RSC (Kubik et al., 2012), but many questions remain methodological issues, such as divergences in experimental open. By combining the latest optogenetic and chemogenetic tech- design or even species specific differences, can explain the dis- niques with electrophysiological recordings and behavioural crepancies in the literature. Nevertheless, this issue can readily be assays, researchers will be able to address questions about the addressed empirically. One approach would be to take a task that nature of functional interactions between RSC and hippocampus is known to induce the expression of IEG within RSC and com- with far greater anatomical and temporal precision. pare the effects of blocking IEG expression with antisense oligo- Future studies could also explore whether RSC and hip- nucleotides at different stages of task acquisition (early versus pocampus are engaged by different navigational strategies, such late stage). Pharmacological or chemogenetic silencing of RSC as map-based, route planning, and scene construction. neuronal activity could similarly be used to assess whether the Furthermore, RSC may work separately from the hippocampus in RSC is differentially involved in remote or recent spatial memory processing previously consolidated spatial information (see (Corcoran et al., 2011). Studies in rodents can be complemented above), as evident in a recent work by Patai et al. (2017). This by imaging studies in humans that compare RSC activity in par- study reported higher RSC activity during distance coding in ticipants navigating in new or previously learnt virtual or real familiar environments, in contrast to higher hippocampal activity environments (Patai et al., 2017). seen in newly learned environments, where more route planning might have occurred. What is the relationship between RSC and hippocampus? What is the role of RSC in episodic memory RSC first attracted attention because of its links with the hip- more broadly? pocampal memory system, and as discussed here, many of the deficits arising from RSC damage resemble those of hippocam- A review of the human literature reveals a difference between left pal lesions, with some notable differences. Of interest is the and right RSC in both lesion findings and imaging; in particular, asymmetric relationship with hippocampus, in that RSC receives the left seems to be more implicated in general episodic memory, more direct connections (from CA1 and subiculum) than it sends, while the right is more implicated in spatial processing. Is RSC although it does project indirectly to hippocampus via entorhinal also involved in episodic memory in animals? We still lack a cortex and the subicular complex. It will thus be important to good animal model of this form of memory, because most animal determine the interaction between these structures, during mem- tasks require training whereas episodes are, by their nature, tran- ory formation, retrieval and updating. sient. Nevertheless it will be important to determine, in future, Targeted combinations of anterograde and retrograde trans- the extent to which RSC has a role in memory that extends ported opsins and optogenetic and chemogenetic interventions beyond space. Indeed, evidence is now emerging implicating 10 Brain and Neuroscience Advances Alexander AS and Nitz DA (2017) Spatially periodic activation patterns RSC in mnemonic processes that do not contain any obvious spa- of retrosplenial cortex encode route sub-spaces and distance trav- tial component including the processing or retrieval of temporal elled. Current Biology 27(11): 1551–1560. information (Powell et al., 2017; Todd et al., 2015) as well as An Y, Varma VR, Varma S, et al. (2017) Evidence for brain glucose dys- learning the inter-relationship between sensory stimuli in the regulation in Alzheimer’s disease. Alzheimer’s & Dementia. Epub environment (Robinson et al., 2014), processes that are likely to ahead of print 19 October. DOI: 10.1016/j.jalz.2017.09.011. be central to our ability to remember an event. Reconciling these Auger SD and Maguire EA (2013) Assessing the mechanism of response seemingly disparate spatial and non-spatial roles, therefore, rep- in the retrosplenial cortex of good and poor navigators. Cortex resents a key challenge for understanding RSC function. 49(10): 2904–2913. Auger SD, Mullally SL and Maguire EA (2012) Retrosplenial cortex codes for permanent landmarks. PLoS ONE 7(8): e43620. Auger SD, Zeidman P and Maguire EA (2015) A central role for the Summary and conclusion retrosplenial cortex in de novo environmental learning. Elife 18(4). In conclusion, we have reviewed the literature on the RSC contri- DOI: 10.7554/eLife.09031. bution to spatial memory and have found that there are three Auger SD, Zeidman P and Maguire EA (2017) Efficacy of navigation may be influenced by retrosplenial cortex-mediated learning of land- broad classes of models which differ in their focus but have sig- mark stability. Neuropsychologia 104: 102–112. nificant overlaps. It remains unclear whether RSC has more than Auger SD and Maguire EA (2018a) Dissociating landmark stability from one function, or whether some overarching model that can orienting Value Using functional magnetic resonance imaging. Jour- explain the current findings better describes these three classes of nal of Cognitive Neuroscience (in press), doi:10.1162/jocn_a_01231 function. We have outlined some open questions, the answers to Auger SD and Maguire EA (2018b) Retrosplenial cortex indexes stability which will require an interaction between multiple different beyond the spatial domain. Journal of Neuroscience, 38(6):1472–1481. approaches, in a variety of species. Baleydier C and Mauguiere F (1985) Anatomical evidence for Over 100 years have passed since Brodmann first identified medial pulvinar connections with the posterior cingulate cortex, the RSC, and, while in the intervening years significant advances the retrosplenial area, and the posterior parahippocampal gyrus in have been made in elucidating the role RSC plays in cognition, monkeys. The Journal of Comparative Neurology 232(2): 219– the precise functions of the RSC still remain somewhat of an Berger TW, Weikart CL, Bassett JL, et al. (1986) Lesions of the ret- enigma. It is hoped that the framework set out in this review will rosplenial cortex produce deficits in reversal learning of the rabbit provide a basis for subsequent endeavours to probe the underly- nictitating membrane response: Implications for potential interac- ing function(s) of this most fascinating of brain structures. tions between hippocampal and cerebellar brain systems. Behavioral Neuroscience 100(6): 802–809. Declaration of conflicting interests Buckley MJ and Mitchell AS (2016) Retrosplenial cortical contributions to anterograde and retrograde memory in the monkey. Cerebral Cor- The author(s) declared no potential conflicts of interest with respect to tex 26(6): 2905–2918. the research, authorship and/or publication of this article. Buckwalter JA, Parvizi J, Morecraft RJ, et al. (2008) Thalamic projec- tions to the posteromedial cortex in the macaque. The Journal of Funding Comparative Neurology 507(5): 1709–1733. This work was supported by a Wellcome Trust Senior Research Bussey TJ, Muir JL, Everitt BJ, et al. (1996) Dissociable effects of anterior Fellowship in Basic Biomedical Sciences to A.S.M. (110157/Z/15/Z), a and posterior cingulate cortex lesions on the acquisition of a condi- grant from National Science Centre (Poland) Sonata Bis 2014//14/E/ tional visual discrimination: Facilitation of early learning vs. impair- NZ4/00172 to R.C., a scholarship from Chinese Scholarship Council to ment of late learning. Behavioural Brain Research 82(1): 45–56. N.Z. (201608000007), a Wellcome Trust Investigator Award to K.J. Byrne P, Becker S and Burgess N (2007) Remembering the past and (WT103896AIA) and A.J.D.N by grants from the BBSRC (BB/ imagining the future: A neural model of spatial memory and imag- H020187/1 and BB/L021005/1). ery. Psychological Review 114(2): 340–375. Cajal SR (1955) Studies on the Cerebral Cortex: Limbic Structures. Chi- cago, IL: Year Book Publishers. ORCID iDs Callaway EM and Luo L (2015) Monosynaptic circuit tracing with glyco- Anna S Mitchell https://orcid.org/0000-0001-8996-1067 protein-deleted rabies viruses. The Journal of Neuroscience 35(24): Kate Jeffery https://orcid.org/0000-0002-9495-0378 8979–8985. Cenquizca LA and Swanson LW (2007) Spatial organization of direct hippocampal field CA1 axonal projections to the rest of the cerebral References cortex. Brain Research Reviews 56(1): 1–26. Aggleton JP, Hunt PR and Rawlins JN (1986) The effects of hippocam- Chen LL, Lin LH, Barnes CA, et al. (1994a) Head-direction cells in pal lesions upon spatial and non-spatial tests of working memory. the rat posterior cortex, II, contributions of visual and ideothetic Behavioural Brain Research 19(2): 133–146. information to the directional firing. Experimental Brain Research Aggleton JP, Hunt PR, Nagle S, et al. (1996) The effects of selective 101(1): 24–34. lesions within the anterior thalamic nuclei on spatial memory in the Chen LL, Lin LH, Green EJ, et al. (1994b) Head-direction cells in the rat rat. Behavioural Brain Research 81(1–2): 189–198. posterior cortex, I, anatomical distribution and behavioral modula- Aggleton JP, Saunders RC, Wright NF, et al. (2014) The origin of pro- tion. Experimental Brain Research 101(1): 8–23. jections from the posterior cingulate and retrosplenial cortices to the Cho J and Sharp PE (2001) Head direction, place, and movement corre- anterior, medial dorsal and laterodorsal thalamic nuclei of macaque lates for cells in the rat retrosplenial cortex. Behavioral Neuroscience monkeys. The European Journal of Neuroscience 39(1): 107–123. 115(1): 3–25. Alexander AS and Nitz DA (2015) Retrosplenial cortex maps the con- Claessen MH and Van der Ham IJ (2017) Classification of navigation junction of internal and external spaces. Nature Neuroscience 18(8): impairment: A systematic review of neuropsychological case stud- 1143–1151. ies. Neuroscience and Biobehavioral Reviews 73: 81–97. Mitchell et al. 11 Clark BJ, Bassett JP, Wang SS, et al. (2010) Impaired head direction cell Iaria G, Chen JK, Guariglia C, et al. (2007) Retrosplenial and hippocam- representation in the anterodorsal thalamus after lesions of the ret- pal brain regions in human navigation: Complementary functional rosplenial cortex. The Journal of Neuroscience 30(15): 5289–5302. contributions to the formation and use of cognitive maps. The Euro- Cooper BG and Mizumori SJ (1999) Retrosplenial cortex inactivation pean Journal of Neuroscience 25(3): 890–899. selectively impairs navigation in darkness. NeuroReport 10(3): Ino T, Doi T, Hirose S, et al. (2007) Directional disorientation follow- 625–630. ing left retrosplenial hemorrhage: A case report with fMRI studies. Cooper BG and Mizumori SJ (2001) Temporary inactivation of the retro- Cortex 43(2): 248–254. splenial cortex causes a transient reorganization of spatial coding in Insausti R, Amaral DG and Cowan WM (1987) The entorhinal cortex the hippocampus. The Journal of Neuroscience 21(11): 3986–4001. of the monkey: II. Cortical afferents. The Journal of Comparative Cooper BG, Manka TF and Mizumori SJ (2001) Finding your way in Neurology 264(3): 356–395. the dark: The retrosplenial cortex contributes to spatial memory and Insausti R and Amaral DG (2008) Entorhinal cortex of the monkey: IV. navigation without visual cues. Behavioral Neuroscience 115(5): Topographical and laminar organization of cortical afferents. The 1012–1028. Journal of Comparative Neurology 509(6): 608–641. Corcoran KA, Donnan MD, Tronson NC, et al. (2011) NMDA recep- Jacob PY, Casali G, Spieser L, et al. (2017) An independent, landmark- tors in retrosplenial cortex are necessary for retrieval of recent and dominated head-direction signal in dysgranular retrosplenial cortex. remote context fear memory. The Journal of Neuroscience 31(32): Nature Neuroscience 20(2): 173–175. 11655–11659. Jeffery KJ, Page HJI and Stringer SM (2016) Optimal cue combination Cowansage KK, Shuman T, Dillingham BCC, et al. (2014) Direct reac- and landmark-stability learning in the head direction system. The tivation of a coherent neocortical memory of context. Neuron 84(2): Journal of Physiology 594(22): 6527–6534. 432–441. Jones BF and Witter MP (2007) Cingulate cortex projections to the para- Czajkowski R, Jayaprakash B, Wiltgen B, et al. (2014) Encoding and hippocampal region and hippocampal formation in the rat. Hippo- storage of spatial information in the retrosplenial cortex. Proceed- campus 17(10): 957–976. ings of the National Academy of Sciences of the United States of Jones BF, Groenewegen HJ and Witter MP (2005) Intrinsic connections America 111(23): 8661–8666. of the cingulate cortex in the rat suggest the existence of multiple Czajkowski R, Sugar J, Zhang SJ, et al. (2013) Superficially projecting functionally segregated networks. Neuroscience 133(1): 193–207. principal neurons in layer v of medial entorhinal cortex in the rat Keene CS and Bucci DJ (2009) Damage to the retrosplenial cortex pro- receive excitatory retrosplenial input. The Journal of Neuroscience duces specific impairments in spatial working memory. Neurobiol- 33(40): 15779–15792. ogy of Learning and Memory 91(4): 408–414. Elduayen C and Save E (2014) The retrosplenial cortex is necessary Kim JH, Park KY, Seo SW, et al. (2007) Reversible verbal and visual for path integration in the dark. Behavioural Brain Research 272: memory deficits after left retrosplenial infarction. Journal of Clinical 303–307. Neurology 3(1): 62–66. Epstein RA, Parker WE and Feiler AM (2007) Where am I now? Distinct Kim M, Jeffery KJ and Maguire EA (2017) Multivoxel pattern analysis roles for parahippocampal and retrosplenial cortices in place recog- reveals 3D place information in the human hippocampus. The Jour- nition. The Journal of Neuroscience 27(23): 6141–6149. nal of Neuroscience 37(16): 4270–4279. Gabriel M (1990) Functions of anterior and posterior cingulate cortex Knight R, Piette CE, Page H, et al. (2014) Weighted cue integration in the during avoidance learning in rabbits. Progress in Brain Research 85: rodent head direction system. Philosophical Transactions of the Royal 467–483. Society of London, Series B, Biological Sciences 369(1635): 20120512. Gaffan D (1994) Scene-specific memory for objects: A model of episodic Kobayashi Y and Amaral DG (2003) Macaque monkey retrosplenial memory impairment in monkeys with fornix transection. Journal of cortex: II, cortical afferents. The Journal of Comparative Neurology Cognitive Neuroscience 6(4): 305–320. 466(1): 48–79. Ghosh VE and Gilboa A (2014) What is a memory schema? A historical Kobayashi Y and Amaral DG (2007) Macaque monkey retrosplenial cor- perspective on current neuroscience literature. Neuropsychologia. tex: III, cortical efferents. The Journal of Comparative Neurology Epub ahead of print 23 November 2013. DOI: 10.1016/j.neuropsy- 502(5): 810–833. chologia.2013.11.010. Kononenko NL and Witter MP (2012) Presubiculum layer III conveys Gilmore AW, Nelson SM, Chen HY, et al. (2017) Task-related and retrosplenial input to the medial entorhinal cortex. Hippocampus resting-state fMRI identify distinct networks that preferentially 22(4): 881–895. support remembering the past and imagining the future. Neuropsy- Kubik S, Miyashita T, Kubik-Zahorodna A, et al. (2012) Loss of activity- chologia. Epub ahead of print 15 June. DOI: 10.1016/j.neuropsycho- dependent arc gene expression in the retrosplenial cortex after hip- logia.2017.06.016. pocampal inactivation: Interaction in a higher-order memory circuit. Guillery RW (1956) Degeneration in the post-commissural fornix Neurobiology of Learning and Memory 97(1): 124–131. and the mamillary peduncle of the rat. Journal of Anatomy 90(3): Laurens J, Kim B, Dickman JD, et al. (2016) Gravity orientation tun- 350–370. ing in macaque anterior thalamus. Nature Neuroscience 19(12): Guillery RW (1957) Degeneration in the hypothalamic connexions of the 1566–1568. albino rat. Journal of Anatomy 91(1): 91–115. Lozano YR, Page HI, Jacob PY, Lomi E, Street J, Jeffery KJ (2017) Ret- Hashimoto R, Tanaka Y and Nakano I (2010) Heading disorientation: A rosplenial and postsubicular head direction cells compared during new test and a possible underlying mechanism. European Neurology visual landmark discrimination. Brain and Neuroscience Advances, 63(2): 87–93. https://doi.org/10.1177/2398212817721859 Hashimoto R, Uechi M, Yumura W, et al. (2016) Egocentric disorien- Maddock RJ (1999) The retrosplenial cortex and emotion: New insights tation and heading disorientation: Evaluation by a new test named from functional neuroimaging of the human brain. Trends in Neuro- card placing test. Rinsho Shinkeigaku = Clinical Neurology 56(12): sciences 22(7): 310–316. 837–845. Maguire EA (2001) The retrosplenial contribution to human navigation: Hindley EL, Nelson AJD, Aggleton JP, et al. (2014) The rat retrosplenial A review of lesion and neuroimaging findings. Scandinavian Jour- cortex is required when visual cues are used flexibly to determine nal of Psychology 42(3): 225–238. location. Behavioural Brain Research 263: 98–107. Mao D, Kandler S, McNaughton BL, et al. (2017) Sparse orthogonal pop- Honda Y and Ishizuka N (2015) Topographic distribution of cortical pro- ulation representation of spatial context in the retrosplenial cortex. jection cells in the rat subiculum. Neuroscience Research 92: 1–20. Nature Communications 8(1): 243. 12 Brain and Neuroscience Advances Marchette SA, Vass LK, Ryan J, et al. (2014) Anchoring the neural com- Pothuizen HHJ, Aggleton JP and Vann SD (2008) Do rats with retrosple- pass: Coding of local spatial reference frames in human medial pari- nial cortex lesions lack direction? The European Journal of Neuro- etal lobe. Nature Neuroscience 17(11): 1598–1606. science 28(12): 2486–2498. Marr D (1971) Simple memory: A theory for archicortex. Philosophical Pothuizen HHJ, Davies M, Albasser MM, et al. (2009) Granular and dys- Transactions of the Royal Society of London, Series B, Biological granular retrosplenial cortices provide qualitatively different contri- Sciences 262(841): 23–81. butions to spatial working memory: Evidence from immediate-early Maviel T, Durkin TP, Menzaghi F, et al. (2004) Sites of neocortical reor- gene imaging in rats. The European Journal of Neuroscience 30(5): ganization critical for remote spatial memory. Science 305(5680): 877–888. 96–99. Powell AL, Vann SD, Olarte-Sánchez CM, et al. (2017) The retrosplenial Meunier M and Destrade C (1988) Electrolytic but not ibotenic acid cortex and object recency memory in the rat. The European Journal lesions of the posterior cingulate cortex produce transitory facilita- of Neuroscience 45(11): 1451–1464. tion of learning in mice. Behavioural Brain Research 27(2):161–172. Powell EW (1978) The cingulate bridge between allocortex, isocortex Minoshima S, Giordani B, Berent S, et al. (1997) Metabolic reduction and thalamus. The Anatomical Record 190(4): 783–7–93. in the posterior cingulate cortex in very early Alzheimer’s disease. Powell TP, Guillery RW and Cowan WM (1957) A quantitative study Annals of Neurology 42(1): 85–94. of the fornixmamillo-thalamic system. Journal of Anatomy 91(4): Mitchell AS and Gaffan D (2008) The magnocellular mediodorsal thala- 419–437. mus is necessary for memory acquisition, but not retrieval. The Jour- Ranganath C and Ritchey M (2012) Two cortical systems for memory- nal of Neuroscience 28(1): 258–263. guided behaviour. Nature Reviews Neuroscience 13(10): 713–726. Mitchell AS, Browning PGF, Wilson CRE, et al. (2008) Dissociable roles Robertson CE, Hermann KL, Mynick A, et al. (2016) Neural repre- for cortical and subcortical structures in memory retrieval and acqui- sentations integrate the current field of view with the remembered sition. The Journal of Neuroscience 28(34): 8387–8396. 360° panorama in scene-selective cortex. Current Biology 26(18): Miyashita T and Rockland KS (2007) GABAergic projections from the 2463–2468. hippocampus to the retrosplenial cortex in the rat. The European Robinson S, Todd TP, Pasternak AR, et al. (2014) Chemogenetic silenc- Journal of Neuroscience 26(5): 1193–1204. ing of neurons in retrosplenial cortex disrupts sensory precondition- Morris, RG (2006) Elements of a neurobiological theory of hippocampal ing. The Journal of Neuroscience 34(33): 10982–10988. function: the role of synaptic plasticity, synaptic tagging and sche- Ruggiero G, Frassinetti F, Iavarone A, et al. (2014) The lost ability to mas. Eur J Neurosci. 23(11):2829-46. find the way: Topographical disorientation after a left brain lesion. Mullally SL, Hassabis D and Maguire EA (2012) Scene construction Neuropsychology 28(1): 147–160. in amnesia: An fMRI study. The Journal of Neuroscience 32(16): Sestieri C, Corbetta M, Romani GL, et al. (2011) Episodic memory 5646–5653. retrieval, parietal cortex, and the default mode network: functional Murray EA and Wise SP (2010) What, if anything, can monkeys tell us and topographic analyses. The Journal of Neuroscience 31(12): about human amnesia when they can’t say anything at all? Neuropsy- 4407–4420. chologia 48(8): 2385–2405. Shine JP, Valdés-Herrera JP, Hegarty M, et al. (2016) The human retro- Neave N, Lloyd S, Sahgal A, et al. (1994) Lack of effect of lesions in splenial cortex and thalamus code head direction in a global refer- the anterior cingulate cortex and retrosplenial cortex on certain tests ence frame. The Journal of Neuroscience 36(24): 6371–6381. of spatial memory in the rat. Behavioural Brain Research 65(1): Shinkai M, Yokofujita J, Oda S, et al. (2005) Dual axonal terminations 89–101. from the retrosplenial and visual association cortices in the laterodorsal Nelson AJD, Hindley EL, Pearce JM, et al. (2015a) The effect of retro- thalamic nucleus of the rat. Anatomy and Embryology 210(4): 317– splenial cortex lesions in rats on incidental and active spatial learn- 326. ing. Frontiers in Behavioral Neuroscience 9: 11. DOI: 10.3389/ Silson EH, Steel AD and Baker CI (2016) Scene-selectivity and retino- fnbeh.2015.00011. eCollection 2015. topy in medial parietal cortex. Frontiers in Human Neuroscience 10: Nelson AJD, Powell AL, Holmes JD, et al. (2015b) What does 412. DOI: 10.3389/fnhum.2016.00412. spatial alternation tell us about retrosplenial cortex function? Smith DM, Barredo J and Mizumori SJY (2012) Complimentary roles Frontiers in Behavioral Neuroscience 9: 126. DOI: 10.3389/ of the hippocampus and retrosplenial cortex in behavioral context fnbeh.2015.00126. discrimination. Hippocampus 22(5): 1121–1133. Nestor PJ, Fryer TD, Ikeda M, et al. (2003) Retrosplenial cortex (BA Smith DM, Freeman JH Jr, Nicholson D, et al. (2002) Limbic thalamic 29/30) hypometabolism in mild cognitive impairment (prodromal lesions, appetitively motivated discrimination learning, and training- Alzheimer’s disease). The European Journal of Neuroscience 18(9): induced neuronal activity in rabbits. The Journal of Neuroscience 2663–2667. 22(18): 8212–8221. Nonaka M, Fitzpatrick R, Lapira J, et al. (2017) Everyday memory: Spiers HJ and Maguire EA (2006) Thoughts, behaviour, and brain Towards a translationally effective method of modelling the encod- dynamics during navigation in the real world. NeuroImage 31(4): ing, forgetting and enhancement of memory. The European Journal 1826–1840. of Neuroscience 46(4): 1937–1953. Spreng RN, Mar RA and Kim ASN (2009) The common neural basis of Parron C and Save E (2004) Comparison of the effects of entorhinal and autobiographical memory, prospection, navigation, theory of mind, retrosplenial cortical lesions on habituation, reaction to spatial and and the default mode: A quantitative meta-analysis. Journal of Cog- non-spatial changes during object exploration in the rat. Neurobiol- nitive Neuroscience 21(3): 489–510. ogy of Learning and Memory 82(1): 1–11. Sripanidkulchai K and Wyss JM (1986) Thalamic projections to retro- Passarelli L, Rosa MGP, Bakola S, et al. (2017) Uniformity and diversity splenial cortex in the rat. The Journal of Comparative Neurology of cortical projections to precuneate areas in the macaque monkey: 254(2): 143–165. What defines area PGm? Cerebral Cortex. Epub ahead of print 25 Sutherland RJ, Whishaw IQ and Kolb B (1988) Contributions of cingu- March. DOI: 10.1093/cercor/bhx067. late cortex to two forms of spatial learning and memory. The Journal Patai EZ, Javadi AH, Ozubko JD, et al. (2017) Long-term consolidation of Neuroscience 8(6): 1863–1872. switches goal proximity coding from hippocampus to retrosplenial Suzuki WA and Amaral DG (1994) Perirhinal and parahippocampal cortex. Available at: http://www.biorxiv.org/content/early/2017/07/ cortices of the macaque monkey: Cortical afferents. The Journal of 25/167882.1.abstract Comparative Neurology 350(4): 497–533. Mitchell et al. 13 Tamir DI and Mitchell JP (2011) The default network distinguishes con- Vann SD and Aggleton JP (2005) Selective dysgranular retrosplenial cor- struals of proximal versus distal events. Journal of Cognitive Neuro- tex lesions in rats disrupt allocentric performance of the radial-arm science 23(10): 2945–2955. maze task. Behavioral Neuroscience 119(6): 1682–1686. Taube JS (1995) Head direction cells recorded in the anterior thalamic Vann SD, Brown MW, Erichsen JT, et al. (2000) Fos imaging reveals nuclei of freely moving rats. The Journal of Neuroscience 15(1 Pt differential patterns of hippocampal and parahippocampal subfield 1): 70–86. activation in rats in response to different spatial memory tests. The Taube JS (2007) The head direction signal: Origins and sensory-motor Journal of Neuroscience 20(7): 2711–2718. integration. Annual Review of Neuroscience 30: 181–207. Vedder LC, Miller AMP, Harrison MB, et al. (2017) Retrosplenial Teipel S, Grothe MJ and Alzheimer’s Disease Neuroimaging Initiative cortical neurons encode navigational cues, trajectories and reward (2016) Does posterior cingulate hypometabolism result from discon- locations during goal directed navigation. Cerebral Cortex 27(7): nection or local pathology across preclinical and clinical stages of 3713–3723. Alzheimer’s disease? European Journal of Nuclear Medicine and Vertes RP, Albo Z, Viana Di and Prisco G (2001) Theta-rhythmically Molecular Imaging 43(3): 526–536. firing neurons in the anterior thalamus: implications for mnemonic Todd TP, Meyer HC and Bucci DJ (2015) Contribution of the retrosple- functions of Papez’s circuit. Neuroscience 104(3): 619–625. nial cortex to temporal discrimination learning. Hippocampus 25(2): Vogt BA (1976) Retrosplenial cortex in the rhesus monkey: A cytoar- 137–141. chitectonic and Golgi study. The Journal of Comparative Neurology Tsanov M, Chah E, Vann SD, et al. (2011) Theta-modulated head direc- 169(1): 63–97. tion cells in the rat anterior thalamus. The Journal of Neuroscience Vogt BA and Miller MW (1983) Cortical connections between rat cingu- 31(26): 9489–9502. late cortex and visual, motor, and postsubicular cortices. The Journal Tse D, Takeuchi T, Kakeyama M, et al. (2011) Schema-dependent gene of Comparative Neurology 216(2): 192–210. activation and memory encoding in neocortex. Science 333(6044): Vogt BA, Pandya DN and Rosene DL (1987) Cingulate cortex of the rhe- 891–895. sus monkey: I. Cytoarchitecture and thalamic afferents. The Journal Van Groen T and Wyss MJ (1990) Connections of the retrosplenial of Comparative Neurology 262(2): 256–270. granular a cortex in the rat. The Journal of Comparative Neurology Vogt BA, Vogt L and Laureys S (2006) Cytology and functionally corre- 300(4): 593–606. lated circuits of human posterior cingulate areas. NeuroImage 29(2): Van Groen T and Wyss MJ (1992a) Connections of the retrosplenial dys- 452–466. granular cortex in the rat. The Journal of Comparative Neurology Wesierska M, Adamska I and Malinowska M (2009) Retrosplenial cor- 315(2): 200–216. tex lesion affected segregation of spatial information in place avoid- Van Groen T and Wyss MJ (1992b) Projections from the laterodorsal ance task in the rat. Neurobiology of Learning and Memory 91(1): nucleus of the thalamus to the limbic and visual cortices in the rat. 41–49. The Journal of Comparative Neurology 324(3): 427–448. Whishaw IQ, Maaswinkel H, Gonzalez CL, et al. (2001) Deficits in Van Groen T and Wyss MJ (2003) Connections of the retrosplenial allothetic and idiothetic spatial behavior in rats with posterior granular B cortex in the rat. The Journal of Comparative Neurology cingulate cortex lesions. Behavioural Brain Research 118(1): 463(3): 249–263. 67–76. Van Hoesen G and Pandya DN (1975) Some connections of the entorhi- Wolbers T and Büchel C (2005) Dissociable retrosplenial and hippocam- nal (area 28) and perirhinal (area 35) cortices of the rhesus monkey, pal contributions to successful formation of survey representations. I, temporal lobe afferents. Brain Research 95(1): 1–24. The Journal of Neuroscience 25(13): 3333–3340. Vann SD, Aggleton JP and Maguire EA (2009) What does the retrosple- Wood ER, Dudchenko PA, Robitsek RJ, et al. (2000) Hippocampal neu- nial cortex do? Nature Reviews Neuroscience 10(11): 792–802. rons encode information about different types of memory episodes Vann SD and Aggleton JP (2002) Extensive cytotoxic lesions of the rat occurring in the same location. Neuron 27(3): 623–633. retrosplenial cortex reveal consistent deficits on tasks that tax allo- Wyss J and Van Groen T (1992) Connections between the retrosplenial centric spatial memory. Behavioral Neuroscience 116(1): 85–94. cortex and the hippocampal formation in the rat: A review. Hippo- Vann SD and Aggleton JP (2004) Testing the importance of the retrosple- campus 2(1): 1–11. nial guidance system: Effects of different sized retrosplenial cortex Yamawaki N, Radulovic J and Shepherd GMG (2016) A corticocortical lesions on heading direction and spatial working memory. Behav- circuit directly links retrosplenial cortex to M2 in the mouse. The ioural Brain Research 155(1): 97–108. Journal of Neuroscience 36(36): 9365–9374. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Brain and Neuroscience Advances SAGE

Retrosplenial cortex and its role in spatial cognition:

Loading next page...
 
/lp/sage/retrosplenial-cortex-and-its-role-in-spatial-cognition-4HvdGeo4PU

References (136)

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
10.1177/2398212818757098
Publisher site
See Article on Publisher Site

Abstract

Retrosplenial cortex is a region within the posterior neocortical system, heavily interconnected with an array of brain networks, both cortical and subcortical, that is, engaged by a myriad of cognitive tasks. Although there is no consensus as to its precise function, evidence from both human and animal studies clearly points to a role in spatial cognition. However, the spatial processing impairments that follow retrosplenial cortex damage are not straightforward to characterise, leading to difficulties in defining the exact nature of its role. In this article, we review this literature and classify the types of ideas that have been put forward into three broad, somewhat overlapping classes: (1) learning of landmark location, stability and permanence; (2) integration between spatial reference frames; and (3) consolidation and retrieval of spatial knowledge (schemas). We evaluate these models and suggest ways to test them, before briefly discussing whether the spatial function may be a subset of a more general function in episodic memory. Keywords Learning, memory, cingulate cortex, primate, hippocampal formation, thalamus, neuroimaging, default mode network, immediate-early genes, electrophysiology Received: 17 September 2017; accepted: 18 December 2017 Introduction Retrosplenial cortex (RSC) has fallen within the scope of mem- classes: first, it is involved in the setting of perceived landmarks ory research for at least 40 years (Vogt, 1976) and yet as Vann into a spatial reference frame for use in orientation (spatial and et al. (2009) pointed out in their recent comprehensive review, directional) as well as evaluation of landmark stability; second, it little was discovered about the structure for the first 90 years after stores and reactivates associations between different processing Brodmann first identified it. Since the early 1990s, a growing modes or reference frames for spatial navigation; and third, it has body of evidence has implicated the RSC variously in spatial a time-limited role in the storage and possibly retrieval of hip- memory, navigation, landmark processing and the sense of direc- pocampal-dependent spatial/episodic memories. We conclude tion, visuospatial imagery and past/future thinking, and episodic with some suggestions about how to further refine, and perhaps memory. Early results were difficult to interpret in the absence of ultimately synthesise, these models. precise neuroanatomical, behavioural, electrophysiological and functional data. However, as a consequence of intense research on the RSC, both across animal models using a variety of meth- ods and also in human neuropsychological and imaging studies, Department of Experimental Psychology, University of Oxford, Oxford, a group of theories is now emerging that highlight the involve- UK ment of the RSC in aspects of cognition that go beyond, yet at the Department of Molecular and Cellular Neurobiology, Nencki Institute of Experimental Biology, Warsaw, Poland same time still underlie, our abilities to process spatial informa- Institute of Behavioural Neuroscience, Division of Psychology and tion and retrieve memories. This review will examine the experi- Language Sciences, University College London, London, UK mental data in light of its contribution to spatial cognition, School of Psychology, Cardiff University, Cardiff, UK beginning with a review of the anatomy and connectivity, fol- lowed by functional investigations based on lesion studies, imag- Corresponding author: ing and electrophysiology, and concluding with evaluation and Anna S. Mitchell, Department of Experimental Psychology, University classification of the main ideas that have emerged. We suggest of Oxford, Tinsley Building, Mansfield Road, Oxford OX1 3SR, UK. that the proposals about RSC function fall into at least three Email: anna.mitchell@psy.ox.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 Figure 1. Schematic of the RSC as seen in midsagittal section and located just posterior to the corpus callosum, in humans, rhesus monkeys and rats. Source: Figure by Jeffery (2017); available at: https://doi.org/10.6084/m9.figshare.5414179.v1 under a CC-BY 4.0 licence. the posterior cingulate cortex in primates (Kobayashi and Amaral, 2003). Cortical connections As shown in Figure 2, neural connections of the RSC from the cortex include the parahippocampal region (postrhinal cortex in rodents) (Suzuki and Amaral, 1994), medial entorhinal cortex (Czajkowski et al., 2013; Insausti and Amaral, 2008; Insausti et al., 1987; Jones and Witter, 2007; Van Hoesen and Pandya, 1975) and cingulate cortex (Jones et al., 2005). RSC receives unidirectional inputs from the CA1 field of the hippocampus (Cenquizca and Swanson, 2007; Miyashita and Rockland, 2007) and from the subiculum (Honda and Ishizuka, 2015; Wyss and Van Groen, 1992). It is also interconnected with the extended hippocampal complex, including the presubiculum, postsubicu- lum and parasubiculum (Kononenko and Witter, 2012; Wyss and Figure 2. A schematic diagram detailing the gross connectivity Van Groen, 1992), visuospatial cortical association areas (mainly of retrosplenial cortex. As depicted in the figure, RSC serves as an medial precuneate gyrus, V4 of the occipital lobes and the dorsal interconnected hub for neocortical, hippocampal, parahippocampal bank of the superior temporal sulcus) (Passarelli et al., 2017) and thalamic regions that are functionally involved in the processing and prefrontal cortex (with the heaviest terminations in the dor- of mammalian perceptions important for direction, location, landmarks solateral prefrontal cortex, frontopolar area 10 and area 11 of the and navigation. Different shading is used for effect only. orbitofrontal cortex); these frontal connections are all recipro- cal. RSC also receives inputs directly from V2 of the occipital lobes. There are also prominent excitatory reciprocal connec- Anatomy and connectivity of RSC tions between RSC and posterior secondary motor cortex – namely M2, that have been recently identified in mice (Yamawaki In human and non-human primates, RSC conforms to the cortical et al., 2016). regions that Brodmann identified as areas 29 and 30, which – along with areas 23 and 31 – form part of the posterior cingulate cortex, lying immediately posterior to the corpus callosum Subcortical connections (Figure 1 – left and middle). Rodents lack areas 23 and 31, and RSC itself is located more dorsally and reaches the brain surface In addition, as shown in Figure 2, RSC has major reciprocal (Figure 1 – right). Its central location makes it pivotally posi- subcortical interactions with the anterior (ATN) and laterodor- tioned to receive information from, and readily influence, many sal thalamic nuclei (Aggleton et al., 2014; Kobayashi and key brain regions responsible for the processing of spatial Amaral, 2003, 2007; Van Groen and Wyss, 1990, 1992a and information. 1992b, 2003; Vogt et al., 1987). While the RSC projections to Typically, structural neural connections have been mainly thalamus mainly arise from layer 6, projections from areas 29 derived from studies in animal models (rodents and non-human and 30 provide different densities of terminal fields in the three primates), while the majority of neural connections studied in subdivisions – anteroventral, anteromedial and anterodorsal – humans have been derived functionally. It is known that in of the ATN (Aggleton et al., 2014). Given that the ATN and both rats and primates, the majority of RSC (RSC granular A laterodorsal thalamic nuclei provide major RSC inputs, it is of and granular B, and RSC dysgranular) connections (up to 78%) interest to establish where these two thalamic structures receive originate in or are received from other parts of RSC and from their inputs. Briefly, the laterodorsal thalamus receives inputs Mitchell et al. 3 from the postsubiculum, visual association cortex and the lat- deficits and found that involvement of the RSC was prominent in eral mamillary bodies (Shinkai et al., 2005; Sripanidkulchai and impairments of landmark processing, particularly when it came Wyss, 1986; Taube, 2007; Vogt and Miller, 1983), while the to reporting distances and directions between known landmarks ATN receives inputs directly from the lateral and medial mamil- or describing the positions of known landmarks or buildings on a lary bodies and from the hippocampal formation/ subicular map. complex. Possibly, the key message transmitted from the lateral Experimental lesion studies in non-human primates can be mamillary bodies to the anterodorsal subdivision of the ATN much more precise, and also bilateral, which has provided new and the laterodorsal thalamus is information about the position insights into RSC function. In rhesus macaque monkeys, damage of the head received from the dorsal tegmental nucleus of to the RSC, which included the most caudal part of the posterior Gudden located in the midbrain (Cajal, 1955; Guillery, 1956, cingulate cortex, selectively impaired the ability to retrieve 1957; Powell et al., 1957; Taube, 1995, 2007). In contrast, we object-in-place scene discriminations that the monkeys had pre- do not yet fully know what information is transmitted to the viously learnt (retrograde memory) (Buckley ad Mitchell, 2016). RSC and cingulate cortex via the anteromedial and anteroven- In these tasks, animals have to learn and remember the location tral subdivisions of the ATN, although theta-modulation (Vertes of a discrete object in a spatial scene. In contrast, these same et al., 2001) and theta-modulated head direction (HD)-signalling animals were able to learn new object-in-place scene discrimina- neurons have been identified in the anteroventral subdivision in tions postoperatively (anterograde memory), so their ability to rats (Tsanov et al., 2011), and gravity-tuned neurons have been organise spatial information appeared to remain intact. However, identified in primate ATN (Laurens et al., 2016). In addition to during new learning that involved a 24-h delay period between the above major connections, there are also lesser connections successive sessions of learning the new set of object-in-place dis- with the mediodorsal thalamus and rodent lateral posterior tha- criminations (i.e. from session 1 to session 2), monkeys with lamic nucleus (Aggleton et al., 2014; Powell, 1978). RSC also RSC damage made more errors than controls during postopera- receives inputs from the intralaminar thalamic nuclei (impor- tive session 2 of new learning only. This selective deficit, which tant for arousal) and primate medial pulvinar (supporting visual was present in all monkeys with RSC damage, comprised a spe- attention) (Baleydier and Mauguiere, 1985; Buckwalter et al., cific impairment in their ability to retrieve these new discrimina- 2008; Vogt et al., 2006). tions which they had seen only 24 h beforehand. The task, In general, the anatomy shows that RSC interacts reciprocally object-in-place scene discriminations, incorporates elements of with many brain regions, consistent with its role, described both spatial (e.g. landmark information) and episodic-like mem- below, in a number of core cognitive competences. In particular, ory (unique object-in-place scene discriminations, with one of it is clear that the RSC interacts with many visual areas of the the objects in each discrimination paired with a reward if it is brain across mammalian species. Of interest is the more unidirec- selected) without being explicitly autobiographical in nature tional relationship with hippocampus and with perirhinal cortex. (Gaffan, 1994; Mitchell et al., 2008; Mitchell and Gaffan, 2008; Murray and Wise, 2010). The novel findings observed in the monkeys’ performance led the authors to conclude that an intact Lesion studies RSC is particularly important for the ability to retrieve informa- tion that has been previously acquired, regardless of whether The literature on pure RSC lesions in humans is sparse and these memories are autobiographical, or episodic (in the pure mostly from unilateral pathology due to the rarity with which sense of what/ where and when), or actively spatial in nature localised infarcts or injury occur to this region, and so most of (Buckley and Mitchell, 2016). Finally, the ability to retrieve this our knowledge of human RSC comes from neuroimaging, which information did not require the monkeys to move around in their we discuss later. Most of the identified lesion-induced deficits environment, although the successful executions of self-gener- appear to involve memory and spatial processing. Maguire ated hand-eye coordinated movements (in order to select the cor- (2001) conducted a comprehensive review of the literature on rect object within the scene on the touchscreen) were necessary. RSC extant at the time and concluded that case studies of RSC Studies involving smaller mammals have proved vital in fur- lesions reveal deficits in episodic memory (memory for life thering our understanding of the contribution of the RSC to cog- events), occurring particularly following left-sided lesions, but nition, as they afford far greater neuroanatomical precision than also consistent reports of topographical disorientation (getting is currently possible in primate studies. An early study by Berger lost), with or without concomitant memory deficits, most of et al. (1986) found that rabbits with RSC lesions could acquire a which followed right-sided lesions. The area that was most con- tone-light discrimination, but were profoundly impaired in sistently involved in the pure disorientation cases was Brodmann reversing it, suggesting a failure to modify a recently established area BA30. Maguire (2001) noted: ‘In every case, the patient was memory. Given the dense interconnections between the RSC and able to recognise the landmarks in their neighbourhoods and the hippocampal spatial system, the majority of subsequent retained a sense of familiarity …’. Despite this, none of the lesion studies have focused on spatial learning. patients were able to find their way in familiar environments, and Some of the early studies into the effects of RSC lesions on all but one were unable to learn new routes. Studies since then spatial tasks produced mixed results. This divergence in findings have confirmed the link between RSC lesions and topographic may be attributable to methodological considerations such as the disorientation, with association of left-sided infarct with memory use of electrolytic or ablation lesions, which destroy fibres of deficits (Kim et al., 2007) and of right-sided lesions with spatial passage and consequently may exaggerate the impact of the RSC impairment (Hashimoto et al., 2010, 2016), although spatial damage, while other studies spared the more caudal aspect of the impairment has also been reported in patients with left-sided RSC, which is now known to be critically involved in spatial lesions (Ino et al., 2007; Ruggiero et al., 2014). Claessen and van memory (Vann and Aggleton, 2002, 2004). Despite these earlier der Ham (2017) conducted a review of lesion-related navigation 4 Brain and Neuroscience Advances controversies, there is now very good evidence that RSC lesions properties of the test environment or the juxtaposition of highly in rodents disrupt spatial memory. Deficits are consistently salient visual cues. Rats learnt the location of the platform either reported on tasks that involve allocentric spatial processing, par- by actively swimming to the platform or passively, by being ticularly when – as with the imaging studies – visual cues are repeatedly placed on the platform location. They were then given needed for orientation (Hindley et al., 2014). Such tasks include a test in which they had to swim to the correct location for the learning the fixed or alternating location of a platform in the first time. RSC-lesioned rats were selectively impaired in the Morris watermaze (Sutherland et al., 1988; Vann and Aggleton, passive condition, indicating that RSC damage did not disrupt 2002, 2004; Whishaw et al., 2001), the radial arm maze (Keene navigation per se, but selectively impaired the ability to switch and Bucci, 2009; Pothuizen et al., 2008; Vann and Aggleton, spatial frames of reference and different spatial viewpoints when 2004) and object-in-place discriminations (Parron and Save, navigating to the platform from a novel position in the environ- 2004). There is some evidence that the RSC dysgranular region ment (Nelson et al., 2015a). Similarly, complete RSC or selective (area 30; see Figure 1 – left) may be particularly important for RSC dysgranular lesions disrupted the ability to recognise the processing allocentric space, as rats with selective RSC dysgran- layout of a room from different viewpoints (Hindley et al., 2014). ular lesions were unable to use distal visual cues to guide spatial Taken together, RSC effects appear to depend on the extent to working memory and relied instead on motor sequence informa- which task performance relies on the retrieval of spatial land- tion (Vann and Aggleton, 2005). Furthermore, deficits have also marks for orientation, or the need to switch between different been found on tasks that require the use of directional informa- spatial strategies or viewpoints. This is in line with the proposal tion (Keene and Bucci, 2009; Pothuizen et al., 2008; Vann and that key aspects of RSC functioning include integration of the Aggleton, 2004) as well as self-motion cues (Elduayen and Save, context in which an event occurs, learning about the significance 2014; Whishaw et al., 2001). In some instances, the involvement of such stimuli or updating representations as new information of RSC has been found to be time-limited: for example, Maviel comes on-line. et al. (2004) found that RSC inactivation in mice disrupted the retrieval of a recent 1-day-old spatial memory but not a remote Brain imaging (positron emission 30-day-old one, while Keene and Bucci (2009) found large impairments on radial maze performance for a 30-s delay relative tomography, functional magnetic to a 5-s delay. Findings such as these, combined with the imme- resonance imaging and immediate- diate-early gene study findings described later, and the primate early gene activation) studies mentioned above, suggest a particular role for RSC when spatial information needs to be retrieved from memory. As outlined above, human, primate and rodent RSC lesion stud- In general, the magnitude of spatial deficits after RSC lesions ies have pointed to a role in spatial processing: complementary tends to be smaller and less striking than the spatial impairments evidence comes from research using metabolic brain imaging, associated with either hippocampal or ATN damage. The most particularly positron emission tomography (PET), functional striking demonstration of this difference is T-maze alternation magnetic resonance imaging (fMRI) and immediate-early gene performance, which is acutely sensitive to both hippocampal and activation (IEG) studies. ATN damage (Aggleton et al., 1986, 1996), but is often spared Human neuroimaging studies have been complicated by the after RSC lesions (Meunier and Destrade, 1988; Neave et al., lack of agreement about exactly which regions belong to RSC 1994; Nelson et al., 2015b; Pothuizen et al., 2008). Indeed, the proper. While the scene-selective posterior and ventral bank of full impact of RSC lesions often only emerges under specific the parieto-occipital sulcus is often referred to as RSC, Silson conditions or when animals are required to shift between differ- et al. (2016) have suggested that the term be reserved for the ent spatial metrics. For example, temporary inactivation of the region within the callosal sulcus extending onto the isthmus of RSC selectively impairs navigation in the dark, but not the light the cingulate gyrus. Such distinctions are relevant for the issue of (Cooper et al., 2001). However, Wesierska et al. (2009) found the specificity of RSC processing, as well as its cross-species that rats with RSC dysgranular lesions could learn to avoid the homology, which is still not fully established. shock zone of a rotating platform if the rotation occurred in the In an early PET study of cerebral glucose metabolism, dark, so darkness per se does not seem to be the problem. The rats Minoshima et al. (1997) found reduced activation in the posterior could also learn to avoid the shock zone if this was defined by cingulate in patients with mild cognitive impairment and early allocentric room cues provided there were no conflicting local Alzheimer’s disease, while Nestor et al. (2003) found that the cues; thus, there was not a straightforward impairment of allo- RSC part of the posterior cingulate, was the most consistently centric cue use either. There was a notable impairment when the hypometabolic region. More recent imaging studies have contin- animals had to disregard the local cues and focus on the room ued to confirm that changes in glucose metabolism in the poste- cues. Thus, as the authors noted, impairments arose when rele- rior cingulate cortex, as well as hippocampal atrophy, are early vant and irrelevant cues needed to be segregated. Similarly, biomarkers for Alzheimer’s disease and are likely present many impairments on both the radial arm maze and T-maze often only years before the clinical symptoms appear (e.g. An et al., 2017; emerge when intra-maze cues are placed in conflict with extra- Teipel et al., 2016). maze cues (Nelson et al., 2015b; Pothuizen et al., 2008; Vann and Since the advent of fMRI in cognitive neuroscience, many Aggleton, 2004). studies have investigated RSC activation as subjects perform A further illustration of the selective nature of RSC lesion- tasks in the scanner. Indeed, RSC is now considered to be part of induced spatial deficits comes from an experiment by Nelson the so-called default mode network, which consists of a set of et al., 2015a in which the location of a submerged platform in a brain structures including medial frontal and medial temporal Morris watermaze was determined by either the geometric lobe regions, lateral and medial parietal areas and the RSC (Vann Mitchell et al. 5 et al., 2009), which are active when subjects are not performing when subjects were shown stationary views of the environment a task in the scanner but rather are lying in the scanner at ‘rest’, and had to make orientation judgements (Shine et al., 2016). or actively simulating a situation (particularly one close in time Furthermore, recent work has also examined RSC activation and space to the present (Tamir and Mitchell, 2011), or when they when participants navigate in a virtual 3D environment (Kim are retrieving a memory (Sestieri et al., 2011)). et al., 2017). Interestingly, in this study, the RSC activation was Cognitive tasks that reliably activate RSC in fMRI studies particularly sensitive to the vertical axis of space, which the include most that have a spatial component, especially when this authors suggest may be supporting processing of gravity, which requires use of the visual environment to retrieve previously is a directional cue in the vertical plane and may be useful for 3D learned information in order to orient. These typically involve navigation. Given that there is evidence for both local and global virtual reality simulations in which subjects navigate, by joystick encoding of direction in RSC, the question arises as to how these or sometimes just by imagination, around a virtual environment, might both be accommodated within the one structure; we return such as a town. In one of the earliest studies, Wolbers and Büchel to this question later. (2005) scanned subjects as they learned a virtual maze-like town While the foregoing studies looked at global spatial environ- and found that RSC activation increased steadily with learning ments, work from the Maguire lab has suggested a role for RSC and paralleled increasing map performance. Similarly, in a study in the processing of individual landmarks. Auger et al. (2012) of London taxi drivers in a virtual environment based on real scanned subjects as they viewed a variety of images with a mix- maps of London (Spiers and Maguire, 2006), RSC activation ture of large and smaller objects and found that RSC was acti- occurred during route planning, spontaneous trajectory changes vated only by the spatially fixed, landmark-like objects, and and confirmation of expectations about the upcoming features of furthermore that the extent of activation correlated with naviga- the outside environment - but not, interestingly, expectation vio- tion ability. In a follow-up study using MVPA, Auger and lations. Another fMRI study confirmed that RSC activity was Maguire (2013) showed that decoding of the number of perma- specifically associated with thoughts of location and orientation, nent landmarks in view was possible, and more so in better navi- as opposed to context familiarity or simple object recognition gators, concluding that RSC, in particular, is concerned with (Epstein et al., 2007). In both studies, the overall pattern of RSC encoding every permanent landmark that is in view. They then activation differed from the one observed for hippocampus (Iaria showed that this RSC permanence encoding also occurred when et al., 2007), with the entire RSC active during both encoding and subjects learned about artificial, abstract landmarks in a feature- retrieval of spatial information. less Fog World (Auger et al., 2015), demonstrating that the RSC A related line of work has investigated the encoding of loca- is involved in new learning of landmarks and their spatial stabil- tion and/or direction by RSC. Marchette et al. (2014) performed ity and also that such learning correlates with navigation ability multi-voxel pattern analysis (MVPA) of fMRI brain activation (Auger et al., 2017). Puzzlingly, however, the involvement of patterns on subjects recalling spatial views from a recently RSC seems better correlated with the stability per se than with learned virtual environment. Because MVPA compares fine- the orientational relevance of the landmarks (Auger and Maguire, grained patterns of activation, it allows inferences to be made 2018a). about whether a subject is discriminating stimuli. The virtual Some meta-analyses of human imaging studies have indicated environment comprised a set of four museums located near each that higher RSC activation occurs when subjects process land- other in a virtual park. RSC activity patterns were similar when mark information (Auger et al., 2012; Auger and Maguire, 2013; subjects faced in similar directions and/or occupied similar loca- Maguire, 2001; Mullally et al., 2012; Spiers and Maguire, 2006) tions within each museum, suggesting that RSC was activating and associate the current panoramic visual scene with memory the same representations of local place and local direction, even (Robertson et al., 2016). Further evidence has revealed that RSC though the environments were separated and oriented differently is activated when subjects retrieve autobiographical memories in global space. Similarity judgement reaction times were faster (Maddock, 1999; Spreng et al., 2009) or engage in future think- for homologous directions or locations, suggesting encoding by ing or imagining (Tamir and Mitchell, 2011), although RSC local features independent of global relationships. However, it appears more engaged with past than future spatial/contextual was not demonstrated that subjects had been able to form global thinking (Gilmore et al., 2017). While the retrieval of autobio- maps of the virtual space (i.e. the reference frame in which the graphical memories, imagining and future thinking may not local spaces were set), so the question remains unanswered about explicitly engage spatial processes, they are nonetheless closely whether RSC is also involved in relating directions within a allied to the spatial functions of RSC and its identified role in global space. retrieval, as they require self-referencing to spatial contexts and Robertson et al. (2016) also found encoding of local land- the updating of spatial representations as events are recalled or marks in a setting in which subjects viewed segments of a 360° imagined based on subjective memories. panorama that either did or did not overlap. RSC activation was Animal models, in particular rodent experiments that engage higher when subjects subsequently viewed isolated scenes from their ability to readily explore their spatial environment, have the overlap condition and judged whether it came from the left or provided imaging evidence across mammalian species that high- the right side of the panorama. A study by Shine et al. (2016) did, lights the importance of the RSC for spatial functioning. One however, find evidence for global heading representation in RSC. particular experimental approach is to study RSC functioning They investigated RSC and thalamus activation in subjects who in the intact rodent brain by investigating the extent, and loca- had learned a virtual environment by walking around with a tion, of the activation of learning-induced immediate-early head-mounted display, which provides vestibular and motor cues genes (IEGs; e.g. Arc, Fos or Zif268) after animals have per- to orientation. They found activation of both structures, which formed a behavioural task. Most of these studies have shown both contain directionally sensitive HD cells (discussed below), increased expression of IEGs in the RSC as a consequence of 6 Brain and Neuroscience Advances spatial learning (Maviel et al., 2004; Vann et al., 2000). One of of landmarks within the room as a whole. Some cells behaved the distinct advantages of this approach is that it allows for far like typical HD cells and fired whenever the animal faced in a greater anatomical precision, for example, revealing subregional particular direction in the global space, while others fired in one or layer-specific differences in RSC activity after animals have direction in one compartment and the opposite direction in the performed a spatial task (Pothuizen et al., 2009). other compartment, as if these cells were more interested in local IEG studies have also revealed the involvement of RSC in direction than global direction. This observation is thus reminis- spatial memory formation. Tse et al. (2011) investigated the two cent of the fMRI experiment by Marchette et al. (2014) discussed IEGs, zif268 and Arc, as rats learned flavour-place pairs; they earlier, in which human subjects showed similar RSC activation found up-regulation of these genes in RSC when animals added patterns in local subspaces independent of their global orienta- two new pairs to the set. A more recent approach has been to tion. Together, these results support the idea that RSC might be combine IEG mapping with chronic in vivo two-photon imaging involved in relating spatial reference frames, with some cells to study the dynamics of Fos fluorescent reporter (FosGFP) in responsible for local orientation and others responsible for the RSC dysgranular cortex during acquisition of the watermaze task bigger picture. (Czajkowski et al., 2014). Higher reporter activity was observed More broadly, the findings concerning HD cells suggest that when animals relied on a set of distal visual cues (allocentric RSC neurons may be integrating landmark information coming strategy), as compared to a simple swimming task with one local from the visual cortex, together with the ongoing HD signal being landmark. Moreover, these observations also revealed a small assembled and maintained by more central in the HD network. population of neurons that were persistently reactivated during Such interaction might depend on the strength and/or reliability of subsequent sessions of the allocentric task. This study showed the sensory input (i.e. landmarks) to RSC and/or the HD system that plasticity occurs within RSC during spatial learning and also (Knight et al., 2014), raising the possibility that RSC directional suggested that this structure is critical for formation of the global neurons have the task of evaluating landmarks and deciding representation. Indeed, in another set of experiments, optogenetic whether they are stable and/or reliable enough to help anchor the reactivation of Fos-expressing neuronal ensembles in mouse sense of direction (Jeffery et al., 2016). RSC led to the replication of context-specific behaviours when The above notwithstanding, only around 10% of RSC neurons the animal was in a safe context, devoid of any features of the seem to be HD neurons, the remainder having more complex fir- original training context (Cowansage et al., 2014). ing correlates. Many of these seem related to the actions the ani- Taken together, these complementary human and animal stud- mal is performing. The first systematic analysis by Chen et al. ies highlight that RSC functioning is involved in spatial learning (1994a, 1994b) reported RSC cells related to body turns in addi- and memory, particularly when environmental cues (landmarks) tion to those with spatial firing characteristics. A subsequent are to be used for re-orientation and perhaps navigation. Studies study found RSC cells with firing significantly correlated with of the time course of RSC involvement suggest a dissociation running speed, location and angular head velocity (Cho and between new learning and memory retrieval/updating. The impli- Sharp, 2001). Similarly, cells that respond to specific combina- cation is that perhaps RSC is less involved in spatial perception per tions of location, direction and movement were reported by se, and more involved with visual memory retrieval and editing. Alexander and Nitz (2015), who recorded RSC neurons as rats ran on two identical ‘W’-shaped tracks located at different places in a room. As well as ordinary HD cells, they found cells encod- Single neuron studies ing conjunctions of local position, global position and left/right turning behaviour. In a later study (Alexander and Nitz, 2017), Researchers typically turn to rodent single-neuron studies to some RSC neurons were found to show firing rate peaks that address fine-grained questions about encoding. Chen et al. recurred periodically as animals ran around the edge of a plus (1994a, 1994b) after conducting the first electrophysiological studies of spatial correlates of rodent RSC reported that around maze – some cells activated once per circumnavigation, some 10% of RSC cells in the rat have the properties of HD cells. twice, some four times and so on. Since the environment had These are cells that fire preferentially when the animal faces in a fourfold symmetry, this observation again suggests a possible particular global direction; cells with these properties are found role in relating local and global spatial reference frames. in a variety of brain regions, and are thought to subserve the However, recurring activation patterns having fourfold symmetry sense of direction (Taube, 2007). RSC head direction cells have were also seen when the animal ran on a ring track, with no local very similar properties to those in other regions, although inter- substructure, so it is possible that the cells were responding to estingly they fire slightly in advance of the actual head direction some type of symmetric feature, such as the corners of the room, (Cho and Sharp, 2001; Lozano et al., 2017). However, 90% of the that was present in the distal room cues. RSC neurons had more complex firing correlates, and no clear In contrast to encoding of route, within which every location hypothesis about overall RSC function emerged. that the animal visits along the full trajectory is represented, oth- A later study by Jacob et al. (2017) similarly found a sub- ers have reported encoding of navigational or behaviourally sig- population of HD cells, in the RSC dysgranular cortex only, the nificant cues (e.g. goal-location coding) by RSC in simpler firing of which was controlled by the local environmental cues linear environments. In a study by Smith et al. (2012), animals independently of the global HD signal. They also – like Chen on a plus maze learned to approach the east arm for reward for et al. – found a further sub-population of directionally tuned cells half of each session and then switched to the west. RSC neurons that showed mixed effects, being influenced both by landmarks developed spatially localised activity patches (‘place fields’) and by the global head direction signal. This experiment took that were sensitive to reward-associated locations, and the num- place in an environment composed of two local sub-compart- ber of place fields substantially increased with experience. ments (two connected rectangles) that had opposite arrangements However, unlike co-recorded hippocampal place cells, which Mitchell et al. 7 produce very focal place fields, RSC place fields were dispersed Landmark processing and sometimes covered the entire arms. One function of RSC The first set of views is that RSC has a specific function in the place fields could be enabling the rats to discriminate two behav- encoding of the spatial and directional characteristics, as well as ioural contexts. stability, of landmarks, independent of their identity. This view A recent study by Mao et al. (2017) reported more hippocam- emerges from such findings as that HD-cell sensitivity to land- pal-like activity in RSC cells, finding spatially localised activity marks is reduced following RSC lesions (Clark et al., 2010), that (i.e. place fields) on a treadmill during movements in head-fixed some RSC directionally tuned cells respond to environmental mice. Locations on the track were marked by tactile cues on the landmarks in preference to the main HD network signal (Jacob travelling belt. As with hippocampal place cells, changes in light et al., 2017), that RSC is active when humans process landmark and reward location cause the cells to alter their firing locations permanence (Auger et al., 2012, 2017; Auger and Maguire, 2013) (remap). These observations support the notion that RSC is sensi- and that lesions to RSC in human subjects cause them to lose the tive to spatially informative cues and contextual changes. ability to use landmarks to orient (Iaria et al., 2007). It is also In addition to place-, cue- and reward-location, Vedder et al. supported by findings that rats with RSC lesions are poor at using (2017) reported conjunctive coding. In a light-cued T-maze task, allocentric spatial cues to navigate (Vann and Aggleton, 2005). RSC neurons increased responsiveness to the light cue, mostly By this view, the function of RSC is to process landmarks as cur- irrespective of left–right position, but they also frequently rently perceived and use them to update an already established responded to location or to reward. Responses involved both spatial framework so that in future they can be used for better increased firing (on responding) and decreased firing (off self-localisation and re-orientation. This viewpoint supposes a responding). Interestingly, responding to the light often slightly particular role for landmarks in the ongoing formation and updat- preceded the onset of the light cue, an anticipatory response also ing of spatial representations and is consistent with the close rela- reported earlier in RSC in rabbits in an associative conditioning tionship of RSC to visual areas as well as to the hippocampal task (Smith et al., 2002). The location-sensitive firing on the stem spatial system. Recently, Auger and Maguire suggested that RSC of the T often distinguished forthcoming left and right turns, so- processing of landmarks may be more to do with permanence and called splitter behaviour also seen in hippocampal place cells stability than orientation relevance, and indeed that RSC’s role in (Wood et al., 2000). Thus, although responding is associated with permanence may extend beyond landmarks to other domains a cue, there seems sometimes to be a supra-sensory component (Auger and Maguire 2018a, 2018b). related to a learned expectation. To summarise, then, the results from electrophysiology stud- ies of RSC neurons provide a mixed picture in which spatial pro- Spatial representations cessing dominates, but the nature of the processing is hard to pin down exactly. It is clear that place and heading are represented, The second view, which could be regarded as an extension of the but so are other variables, and the nature and function of the con- first to information beyond landmarks, is that this area serves to junctive encoding remains to be elucidated. mediate between spatial representations, as detailed in the review by Vann et al. (2009). For some investigators, this has meant between egocentric and allocentric processing, although The RSC contribution to spatial egocentric suggests different things to different researchers, meaning self-motion-updated to some and viewpoint-dependent cognition – consensus and to others. Chen et al. (1994a) made the specific proposal that the controversies egocentric information processed by RSC concerns self-motion, The experimental literature reviewed above has revealed areas a position also taken by Alexander and Nitz (2015); this is sup- where investigators are in general agreement, and other areas ported by their and others’ observations that directionally tuned where there is debate or uncertainty. In this section we review neurons in RSC are updated by self-motion (sometimes called these areas and outline some ways forward to resolve these. idiothetic) cues and that navigation in RSC-lesioned animals is There seems to be general agreement that RSC has a role in affected by darkness (Cooper and Mizumori, 1999). Other allocentric spatial processing, as highlighted in the review of authors have taken “egocentric” more broadly to mean spatial Vann et al. (2009), but there are differences in opinion as to the items encoded with respect to the body versus with respect to the exact contribution it makes, and also in whether it has a broader world. For example, Burgess and colleagues have suggested that role in memory, of which space is just a subcomponent. In this RSC is part of the progressive cortical transform of parietal ego- regard, the status of RSC research is a little reminiscent of hip- centrically to hippocampal allocentrically encoded information pocampal research 30 years ago. Our conclusion is that the litera- (Byrne et al., 2007): the hypothesis of egocentric-allocentric ture has yielded three broad, somewhat related views concerning transformation by RSC recurs repeatedly in the literature (see RSC’s spatial function, which we explore further below: Vann et al. (2009) for discussion). Another, not dissimilar view is that RSC is involved in con- 1. It processes landmarks and landmark stability/permanence, structing and relating allocentric spatial reference frames more possibly in service of spatial/directional orientation or per- generally, not necessarily egocentric/allocentric ones. This haps more broadly. view is supported by studies such as the museums study of 2. It mediates between spatial representations, processing Marchette et al. (2014), which found similarities in the encod- modes or reference frames. ing of local spaces even though these were separated in global 3. It is involved in consolidation and retrieval of spatial space, and the similar findings of Jacob et al. (2017) that some schemas, for example to support episodic memory. RSC neurons constructed a directional signal based on local 8 Brain and Neuroscience Advances cues while others used the global space. A related notion is that compares these external cues with internal models of the situa- RSC is involved in switching between different modes (as tion, including input regarding self-motion. opposed to frames) of spatial processing, such as from light to dark (Cooper et al., 2001; Cooper and Mizumori, 1999), or dis- Open questions tal to proximal cues and so on. These models all share the underlying feature that RSC has Resolving the above ideas into a single, inclusive model of RSC access to the same spatial information represented in different function (if this is possible) will require the answering of some of ways, and is needed in order to switch between these. the outstanding questions raised by the studies to date. Below, we outline some of these questions. Spatial schema consolidation/retrieval Does RSC have a specific interest in The final view, which is broadest of all, is that RSC is involved in formation and consolidation of hippocampus-dependent spatial/ landmarks, as a subclass of spatial cue? episodic memories. What differentiates these models from the A finding that has emerged from multiple studies of spatial pro- foregoing, and also from standard theories of hippocampal function, cessing is that RSC is particularly involved in the processing of is the incorporation of a temporal dimension to the encoding. By landmarks, which is to say discrete objects or visual discontinui- this is meant that RSC is not needed for de novo spatial learning, ties in the panorama that serve, by virtue of their distant location but is required when the animal needs to draw on a previously and spatial stability, to orient the sense of direction. However, the learned set of spatial relationships, in order to execute a task or specific hypothesis that it is interested in landmarks as discrete acquire new information to add to its stored representation. objects as opposed to, say, visual panoramas, has not been fully These ideas draw on two sets of theoretical work already tested. An unanswered question then is whether RSC is engaged extant in the literature: the idea proposed by Marr (1971) that rap- during spatial processing in the absence of landmarks; for exam- idly formed hippocampal memories are slowly consolidated in ple, in an environment devoid of discrete spatial cues in which neocortex, and the idea that spatial learning entails the formation geometry or smooth visual shading provides the only cues to not just of task- or item-specific memories, but also of a more direction. It should be noted that most types of geometric envi- general framework within which the memories are situated, which ronment (squares, rectangles, teardrops, etc) have corners, which has sometimes been called a schema (Morris, 2006; Ghosh and could in principle act as discrete landmarks, so care would have Gilboa, 2014). An example of a schema might be the watermaze to be taken with the environmental design to ensure the absence task, in which a rat is faced with needing to learn a new platform of all such discrete visual stimuli. The general question to be location: learning of the new location is faster than the original answered here is whether the brain, via the RSC, treats landmarks because the rat already knows the layout of the room and the as a special category of object or whether the interest of RSC in watermaze, and the procedures required to learn the platform landmarks stems solely from their spatial utility, derived from the location – learning the location for today requires just a small constant spatial relations between them, or from their permanent updating. nature, irrespective of their status as landmarks (Auger and Support for this consolidation/updating idea comes from mul- Maguire, 2018 a and b). tiple observations in the literature that the role of RSC in behav- ioural tasks is frequently time-limited in that its effects occur later in training rather than immediately. In particular, there Does RSC mediate between spatial seems to be a 24-h time window after training, below which RSC representations? is engaged less, but after which it becomes involved: see the The core idea here is that RSC may not be needed for spatial experiment by Buckley and Mitchell (2016), and the observation learning per se, but is needed when the subject moves between by Bontempi and colleagues that IEGs are up-regulated when representational modes. This may entail switching from egocen- mice retrieve a 30-day-old memory, but not a 1-day-old one tric to allocentric encoding of cues, or relating an interior space (Maviel et al., 2004). IEG studies have also revealed greater to an exterior one (e.g. deciding which door one needs to exit engagement for RSC in spaced versus massed training (Nonaka through to reach the carpark). This view is an extension of the et al., 2017), consistent with the need, in spaced training, for local idea discussed above, that RSC is needed to be able to use reactivation of a partly consolidated (10-min-old) memory rather spatial landmarks to retrieve current location and heading. The than a completely newly formed (30-s-old) one. important new ingredient supplied by the reference frame frame- This time-dependency has led several investigators to propose work, as it were, is that at least two representations have had to that RSC is part of a primacy system (Bussey et al., 1996; Gabriel, be activated: for example, being in one place and thinking about 1990), the function of which is to retrieve and process informa- another, or navigating in the dark and remembering where things tion learned earlier. Ranganath and Ritchey (2012) put forward are based on experience in the light. The question to be answered, one of the most detailed of such models that proposed that RSC, therefore, is whether RSC is indeed needed for a subject to acti- together with parahippocampal cortex, form part of a posterior vate two representations simultaneously. cortical network that functions to support episodic memory. They Testing this idea is complicated by the demonstration dis- suggest that this network matches incoming cues about the cur- cussed earlier that RSC is also needed to use local spatial cues to rent behavioural context to what they call situation models, retrieve a previously learned spatial layout. Since it is required which are internally stored representations of the relationships for current self-localisation, the idea that it is also needed for among the entities and the environment. According to their view, spatial imagination, or route planning, or future thinking, or other the parahippocampal cortex identifies contexts and the RSC Mitchell et al. 9 similar imagination-based functions, is hard to test directly will be useful in these studies, as they allow more precise inter- because a lesioned subject can’t even get past the initial orienta- ruption of selective neurons. Given the well-known connectivity tion problem. However, more temporally focused interventions of these structures, several experimental schemes can be pro- such as optogenetic (in)activation may be of help here. For exam- posed. For example, the general population of hippocampal neu- ple, once intact animals have learned a radial maze task, it may be rons sending projections to the RSC could be targeted by both found that RSC is needed if the lights are turned off halfway chemogenetic and optogenetic approaches with a retrogradely through a trial, forcing a switch from one processing mode to transported vector. Since the effective illumination of the entire another. Similarly, rats familiarised with a small space inside a hippocampus would be technically challenging given its subcor- larger one may be able to navigate between the two when RSC is tical location, the chemogenetic approach would be currently operating, but not when it is inactivated. These types of task preferred in this case. The presynaptic terminals of the CA1 pro- probe retrieval and manipulation of already-stored spatial jection neurons could be optically activated within RSC after representations. hippocampal vector injections. In this case, an implantable, light- emitting diode (LED)-based module could be positioned on top of dysgranular RSC to stimulate areas 30 and 29c. With the What is the time course of RSC involvement development of efficient orange LEDs, a similar approach could in spatial learning? be used for inhibition. Red-shifted opsins could further increase the range, potentially allowing the illumination of the entire RSC It remains an open question exactly when RSC comes into play and enabling optogenetic intervention in the hippocampus. during formation and use of spatial memories. IEG studies have Finally, transsynaptic circuit labelling with rabies virus could found that activation occurs rapidly (within minutes) of modify- single-out even more specific sub-populations of projection neu- ing a spatial schema (Tse et al., 2011), and the massed/spaced rons (Callaway and Luo, 2015). In all cases, temporally precise learning experiment from the same group suggests a role for RSC interruption of processing epochs could be achieved. during at least the first few minutes (Nonaka et al., 2017). The extent to which RSC and hippocampus are functionally However other animal studies have found that RSC dependence coupled could also be examined by combining temporary modula- of a task does not appear until memories are reactivated again tion techniques with electrophysiology: for example, assessing the after a delay of up to 24 h (e.g. Buckley and Mitchell, 2016), effect of hippocampal inactivation on neuronal firing within RSC presumably to allow an epoch of time to pass before re-engage- or vice versa. Extant data suggests that RSC inactivation causes ment of the RSC memory can occur. Consequently, as our current changes in hippocampal place fields (Cooper and Mizumori, 2001) understanding lies it is difficult to distinguish if there is a critical and the hippocampal inactivation alters experience-dependent time window in which RSC is engaged by spatial tasks or whether plasticity in RSC (Kubik et al., 2012), but many questions remain methodological issues, such as divergences in experimental open. By combining the latest optogenetic and chemogenetic tech- design or even species specific differences, can explain the dis- niques with electrophysiological recordings and behavioural crepancies in the literature. Nevertheless, this issue can readily be assays, researchers will be able to address questions about the addressed empirically. One approach would be to take a task that nature of functional interactions between RSC and hippocampus is known to induce the expression of IEG within RSC and com- with far greater anatomical and temporal precision. pare the effects of blocking IEG expression with antisense oligo- Future studies could also explore whether RSC and hip- nucleotides at different stages of task acquisition (early versus pocampus are engaged by different navigational strategies, such late stage). Pharmacological or chemogenetic silencing of RSC as map-based, route planning, and scene construction. neuronal activity could similarly be used to assess whether the Furthermore, RSC may work separately from the hippocampus in RSC is differentially involved in remote or recent spatial memory processing previously consolidated spatial information (see (Corcoran et al., 2011). Studies in rodents can be complemented above), as evident in a recent work by Patai et al. (2017). This by imaging studies in humans that compare RSC activity in par- study reported higher RSC activity during distance coding in ticipants navigating in new or previously learnt virtual or real familiar environments, in contrast to higher hippocampal activity environments (Patai et al., 2017). seen in newly learned environments, where more route planning might have occurred. What is the relationship between RSC and hippocampus? What is the role of RSC in episodic memory RSC first attracted attention because of its links with the hip- more broadly? pocampal memory system, and as discussed here, many of the deficits arising from RSC damage resemble those of hippocam- A review of the human literature reveals a difference between left pal lesions, with some notable differences. Of interest is the and right RSC in both lesion findings and imaging; in particular, asymmetric relationship with hippocampus, in that RSC receives the left seems to be more implicated in general episodic memory, more direct connections (from CA1 and subiculum) than it sends, while the right is more implicated in spatial processing. Is RSC although it does project indirectly to hippocampus via entorhinal also involved in episodic memory in animals? We still lack a cortex and the subicular complex. It will thus be important to good animal model of this form of memory, because most animal determine the interaction between these structures, during mem- tasks require training whereas episodes are, by their nature, tran- ory formation, retrieval and updating. sient. Nevertheless it will be important to determine, in future, Targeted combinations of anterograde and retrograde trans- the extent to which RSC has a role in memory that extends ported opsins and optogenetic and chemogenetic interventions beyond space. Indeed, evidence is now emerging implicating 10 Brain and Neuroscience Advances Alexander AS and Nitz DA (2017) Spatially periodic activation patterns RSC in mnemonic processes that do not contain any obvious spa- of retrosplenial cortex encode route sub-spaces and distance trav- tial component including the processing or retrieval of temporal elled. Current Biology 27(11): 1551–1560. information (Powell et al., 2017; Todd et al., 2015) as well as An Y, Varma VR, Varma S, et al. (2017) Evidence for brain glucose dys- learning the inter-relationship between sensory stimuli in the regulation in Alzheimer’s disease. Alzheimer’s & Dementia. Epub environment (Robinson et al., 2014), processes that are likely to ahead of print 19 October. DOI: 10.1016/j.jalz.2017.09.011. be central to our ability to remember an event. Reconciling these Auger SD and Maguire EA (2013) Assessing the mechanism of response seemingly disparate spatial and non-spatial roles, therefore, rep- in the retrosplenial cortex of good and poor navigators. Cortex resents a key challenge for understanding RSC function. 49(10): 2904–2913. Auger SD, Mullally SL and Maguire EA (2012) Retrosplenial cortex codes for permanent landmarks. PLoS ONE 7(8): e43620. Auger SD, Zeidman P and Maguire EA (2015) A central role for the Summary and conclusion retrosplenial cortex in de novo environmental learning. Elife 18(4). In conclusion, we have reviewed the literature on the RSC contri- DOI: 10.7554/eLife.09031. bution to spatial memory and have found that there are three Auger SD, Zeidman P and Maguire EA (2017) Efficacy of navigation may be influenced by retrosplenial cortex-mediated learning of land- broad classes of models which differ in their focus but have sig- mark stability. Neuropsychologia 104: 102–112. nificant overlaps. It remains unclear whether RSC has more than Auger SD and Maguire EA (2018a) Dissociating landmark stability from one function, or whether some overarching model that can orienting Value Using functional magnetic resonance imaging. Jour- explain the current findings better describes these three classes of nal of Cognitive Neuroscience (in press), doi:10.1162/jocn_a_01231 function. We have outlined some open questions, the answers to Auger SD and Maguire EA (2018b) Retrosplenial cortex indexes stability which will require an interaction between multiple different beyond the spatial domain. Journal of Neuroscience, 38(6):1472–1481. approaches, in a variety of species. Baleydier C and Mauguiere F (1985) Anatomical evidence for Over 100 years have passed since Brodmann first identified medial pulvinar connections with the posterior cingulate cortex, the RSC, and, while in the intervening years significant advances the retrosplenial area, and the posterior parahippocampal gyrus in have been made in elucidating the role RSC plays in cognition, monkeys. The Journal of Comparative Neurology 232(2): 219– the precise functions of the RSC still remain somewhat of an Berger TW, Weikart CL, Bassett JL, et al. (1986) Lesions of the ret- enigma. It is hoped that the framework set out in this review will rosplenial cortex produce deficits in reversal learning of the rabbit provide a basis for subsequent endeavours to probe the underly- nictitating membrane response: Implications for potential interac- ing function(s) of this most fascinating of brain structures. tions between hippocampal and cerebellar brain systems. Behavioral Neuroscience 100(6): 802–809. Declaration of conflicting interests Buckley MJ and Mitchell AS (2016) Retrosplenial cortical contributions to anterograde and retrograde memory in the monkey. Cerebral Cor- The author(s) declared no potential conflicts of interest with respect to tex 26(6): 2905–2918. the research, authorship and/or publication of this article. Buckwalter JA, Parvizi J, Morecraft RJ, et al. (2008) Thalamic projec- tions to the posteromedial cortex in the macaque. The Journal of Funding Comparative Neurology 507(5): 1709–1733. This work was supported by a Wellcome Trust Senior Research Bussey TJ, Muir JL, Everitt BJ, et al. (1996) Dissociable effects of anterior Fellowship in Basic Biomedical Sciences to A.S.M. (110157/Z/15/Z), a and posterior cingulate cortex lesions on the acquisition of a condi- grant from National Science Centre (Poland) Sonata Bis 2014//14/E/ tional visual discrimination: Facilitation of early learning vs. impair- NZ4/00172 to R.C., a scholarship from Chinese Scholarship Council to ment of late learning. Behavioural Brain Research 82(1): 45–56. N.Z. (201608000007), a Wellcome Trust Investigator Award to K.J. Byrne P, Becker S and Burgess N (2007) Remembering the past and (WT103896AIA) and A.J.D.N by grants from the BBSRC (BB/ imagining the future: A neural model of spatial memory and imag- H020187/1 and BB/L021005/1). ery. Psychological Review 114(2): 340–375. Cajal SR (1955) Studies on the Cerebral Cortex: Limbic Structures. Chi- cago, IL: Year Book Publishers. ORCID iDs Callaway EM and Luo L (2015) Monosynaptic circuit tracing with glyco- Anna S Mitchell https://orcid.org/0000-0001-8996-1067 protein-deleted rabies viruses. The Journal of Neuroscience 35(24): Kate Jeffery https://orcid.org/0000-0002-9495-0378 8979–8985. Cenquizca LA and Swanson LW (2007) Spatial organization of direct hippocampal field CA1 axonal projections to the rest of the cerebral References cortex. Brain Research Reviews 56(1): 1–26. Aggleton JP, Hunt PR and Rawlins JN (1986) The effects of hippocam- Chen LL, Lin LH, Barnes CA, et al. (1994a) Head-direction cells in pal lesions upon spatial and non-spatial tests of working memory. the rat posterior cortex, II, contributions of visual and ideothetic Behavioural Brain Research 19(2): 133–146. information to the directional firing. Experimental Brain Research Aggleton JP, Hunt PR, Nagle S, et al. (1996) The effects of selective 101(1): 24–34. lesions within the anterior thalamic nuclei on spatial memory in the Chen LL, Lin LH, Green EJ, et al. (1994b) Head-direction cells in the rat rat. Behavioural Brain Research 81(1–2): 189–198. posterior cortex, I, anatomical distribution and behavioral modula- Aggleton JP, Saunders RC, Wright NF, et al. (2014) The origin of pro- tion. Experimental Brain Research 101(1): 8–23. jections from the posterior cingulate and retrosplenial cortices to the Cho J and Sharp PE (2001) Head direction, place, and movement corre- anterior, medial dorsal and laterodorsal thalamic nuclei of macaque lates for cells in the rat retrosplenial cortex. Behavioral Neuroscience monkeys. The European Journal of Neuroscience 39(1): 107–123. 115(1): 3–25. Alexander AS and Nitz DA (2015) Retrosplenial cortex maps the con- Claessen MH and Van der Ham IJ (2017) Classification of navigation junction of internal and external spaces. Nature Neuroscience 18(8): impairment: A systematic review of neuropsychological case stud- 1143–1151. ies. Neuroscience and Biobehavioral Reviews 73: 81–97. Mitchell et al. 11 Clark BJ, Bassett JP, Wang SS, et al. (2010) Impaired head direction cell Iaria G, Chen JK, Guariglia C, et al. (2007) Retrosplenial and hippocam- representation in the anterodorsal thalamus after lesions of the ret- pal brain regions in human navigation: Complementary functional rosplenial cortex. The Journal of Neuroscience 30(15): 5289–5302. contributions to the formation and use of cognitive maps. The Euro- Cooper BG and Mizumori SJ (1999) Retrosplenial cortex inactivation pean Journal of Neuroscience 25(3): 890–899. selectively impairs navigation in darkness. NeuroReport 10(3): Ino T, Doi T, Hirose S, et al. (2007) Directional disorientation follow- 625–630. ing left retrosplenial hemorrhage: A case report with fMRI studies. Cooper BG and Mizumori SJ (2001) Temporary inactivation of the retro- Cortex 43(2): 248–254. splenial cortex causes a transient reorganization of spatial coding in Insausti R, Amaral DG and Cowan WM (1987) The entorhinal cortex the hippocampus. The Journal of Neuroscience 21(11): 3986–4001. of the monkey: II. Cortical afferents. The Journal of Comparative Cooper BG, Manka TF and Mizumori SJ (2001) Finding your way in Neurology 264(3): 356–395. the dark: The retrosplenial cortex contributes to spatial memory and Insausti R and Amaral DG (2008) Entorhinal cortex of the monkey: IV. navigation without visual cues. Behavioral Neuroscience 115(5): Topographical and laminar organization of cortical afferents. The 1012–1028. Journal of Comparative Neurology 509(6): 608–641. Corcoran KA, Donnan MD, Tronson NC, et al. (2011) NMDA recep- Jacob PY, Casali G, Spieser L, et al. (2017) An independent, landmark- tors in retrosplenial cortex are necessary for retrieval of recent and dominated head-direction signal in dysgranular retrosplenial cortex. remote context fear memory. The Journal of Neuroscience 31(32): Nature Neuroscience 20(2): 173–175. 11655–11659. Jeffery KJ, Page HJI and Stringer SM (2016) Optimal cue combination Cowansage KK, Shuman T, Dillingham BCC, et al. (2014) Direct reac- and landmark-stability learning in the head direction system. The tivation of a coherent neocortical memory of context. Neuron 84(2): Journal of Physiology 594(22): 6527–6534. 432–441. Jones BF and Witter MP (2007) Cingulate cortex projections to the para- Czajkowski R, Jayaprakash B, Wiltgen B, et al. (2014) Encoding and hippocampal region and hippocampal formation in the rat. Hippo- storage of spatial information in the retrosplenial cortex. Proceed- campus 17(10): 957–976. ings of the National Academy of Sciences of the United States of Jones BF, Groenewegen HJ and Witter MP (2005) Intrinsic connections America 111(23): 8661–8666. of the cingulate cortex in the rat suggest the existence of multiple Czajkowski R, Sugar J, Zhang SJ, et al. (2013) Superficially projecting functionally segregated networks. Neuroscience 133(1): 193–207. principal neurons in layer v of medial entorhinal cortex in the rat Keene CS and Bucci DJ (2009) Damage to the retrosplenial cortex pro- receive excitatory retrosplenial input. The Journal of Neuroscience duces specific impairments in spatial working memory. Neurobiol- 33(40): 15779–15792. ogy of Learning and Memory 91(4): 408–414. Elduayen C and Save E (2014) The retrosplenial cortex is necessary Kim JH, Park KY, Seo SW, et al. (2007) Reversible verbal and visual for path integration in the dark. Behavioural Brain Research 272: memory deficits after left retrosplenial infarction. Journal of Clinical 303–307. Neurology 3(1): 62–66. Epstein RA, Parker WE and Feiler AM (2007) Where am I now? Distinct Kim M, Jeffery KJ and Maguire EA (2017) Multivoxel pattern analysis roles for parahippocampal and retrosplenial cortices in place recog- reveals 3D place information in the human hippocampus. The Jour- nition. The Journal of Neuroscience 27(23): 6141–6149. nal of Neuroscience 37(16): 4270–4279. Gabriel M (1990) Functions of anterior and posterior cingulate cortex Knight R, Piette CE, Page H, et al. (2014) Weighted cue integration in the during avoidance learning in rabbits. Progress in Brain Research 85: rodent head direction system. Philosophical Transactions of the Royal 467–483. Society of London, Series B, Biological Sciences 369(1635): 20120512. Gaffan D (1994) Scene-specific memory for objects: A model of episodic Kobayashi Y and Amaral DG (2003) Macaque monkey retrosplenial memory impairment in monkeys with fornix transection. Journal of cortex: II, cortical afferents. The Journal of Comparative Neurology Cognitive Neuroscience 6(4): 305–320. 466(1): 48–79. Ghosh VE and Gilboa A (2014) What is a memory schema? A historical Kobayashi Y and Amaral DG (2007) Macaque monkey retrosplenial cor- perspective on current neuroscience literature. Neuropsychologia. tex: III, cortical efferents. The Journal of Comparative Neurology Epub ahead of print 23 November 2013. DOI: 10.1016/j.neuropsy- 502(5): 810–833. chologia.2013.11.010. Kononenko NL and Witter MP (2012) Presubiculum layer III conveys Gilmore AW, Nelson SM, Chen HY, et al. (2017) Task-related and retrosplenial input to the medial entorhinal cortex. Hippocampus resting-state fMRI identify distinct networks that preferentially 22(4): 881–895. support remembering the past and imagining the future. Neuropsy- Kubik S, Miyashita T, Kubik-Zahorodna A, et al. (2012) Loss of activity- chologia. Epub ahead of print 15 June. DOI: 10.1016/j.neuropsycho- dependent arc gene expression in the retrosplenial cortex after hip- logia.2017.06.016. pocampal inactivation: Interaction in a higher-order memory circuit. Guillery RW (1956) Degeneration in the post-commissural fornix Neurobiology of Learning and Memory 97(1): 124–131. and the mamillary peduncle of the rat. Journal of Anatomy 90(3): Laurens J, Kim B, Dickman JD, et al. (2016) Gravity orientation tun- 350–370. ing in macaque anterior thalamus. Nature Neuroscience 19(12): Guillery RW (1957) Degeneration in the hypothalamic connexions of the 1566–1568. albino rat. Journal of Anatomy 91(1): 91–115. Lozano YR, Page HI, Jacob PY, Lomi E, Street J, Jeffery KJ (2017) Ret- Hashimoto R, Tanaka Y and Nakano I (2010) Heading disorientation: A rosplenial and postsubicular head direction cells compared during new test and a possible underlying mechanism. European Neurology visual landmark discrimination. Brain and Neuroscience Advances, 63(2): 87–93. https://doi.org/10.1177/2398212817721859 Hashimoto R, Uechi M, Yumura W, et al. (2016) Egocentric disorien- Maddock RJ (1999) The retrosplenial cortex and emotion: New insights tation and heading disorientation: Evaluation by a new test named from functional neuroimaging of the human brain. Trends in Neuro- card placing test. Rinsho Shinkeigaku = Clinical Neurology 56(12): sciences 22(7): 310–316. 837–845. Maguire EA (2001) The retrosplenial contribution to human navigation: Hindley EL, Nelson AJD, Aggleton JP, et al. (2014) The rat retrosplenial A review of lesion and neuroimaging findings. Scandinavian Jour- cortex is required when visual cues are used flexibly to determine nal of Psychology 42(3): 225–238. location. Behavioural Brain Research 263: 98–107. Mao D, Kandler S, McNaughton BL, et al. (2017) Sparse orthogonal pop- Honda Y and Ishizuka N (2015) Topographic distribution of cortical pro- ulation representation of spatial context in the retrosplenial cortex. jection cells in the rat subiculum. Neuroscience Research 92: 1–20. Nature Communications 8(1): 243. 12 Brain and Neuroscience Advances Marchette SA, Vass LK, Ryan J, et al. (2014) Anchoring the neural com- Pothuizen HHJ, Aggleton JP and Vann SD (2008) Do rats with retrosple- pass: Coding of local spatial reference frames in human medial pari- nial cortex lesions lack direction? The European Journal of Neuro- etal lobe. Nature Neuroscience 17(11): 1598–1606. science 28(12): 2486–2498. Marr D (1971) Simple memory: A theory for archicortex. Philosophical Pothuizen HHJ, Davies M, Albasser MM, et al. (2009) Granular and dys- Transactions of the Royal Society of London, Series B, Biological granular retrosplenial cortices provide qualitatively different contri- Sciences 262(841): 23–81. butions to spatial working memory: Evidence from immediate-early Maviel T, Durkin TP, Menzaghi F, et al. (2004) Sites of neocortical reor- gene imaging in rats. The European Journal of Neuroscience 30(5): ganization critical for remote spatial memory. Science 305(5680): 877–888. 96–99. Powell AL, Vann SD, Olarte-Sánchez CM, et al. (2017) The retrosplenial Meunier M and Destrade C (1988) Electrolytic but not ibotenic acid cortex and object recency memory in the rat. The European Journal lesions of the posterior cingulate cortex produce transitory facilita- of Neuroscience 45(11): 1451–1464. tion of learning in mice. Behavioural Brain Research 27(2):161–172. Powell EW (1978) The cingulate bridge between allocortex, isocortex Minoshima S, Giordani B, Berent S, et al. (1997) Metabolic reduction and thalamus. The Anatomical Record 190(4): 783–7–93. in the posterior cingulate cortex in very early Alzheimer’s disease. Powell TP, Guillery RW and Cowan WM (1957) A quantitative study Annals of Neurology 42(1): 85–94. of the fornixmamillo-thalamic system. Journal of Anatomy 91(4): Mitchell AS and Gaffan D (2008) The magnocellular mediodorsal thala- 419–437. mus is necessary for memory acquisition, but not retrieval. The Jour- Ranganath C and Ritchey M (2012) Two cortical systems for memory- nal of Neuroscience 28(1): 258–263. guided behaviour. Nature Reviews Neuroscience 13(10): 713–726. Mitchell AS, Browning PGF, Wilson CRE, et al. (2008) Dissociable roles Robertson CE, Hermann KL, Mynick A, et al. (2016) Neural repre- for cortical and subcortical structures in memory retrieval and acqui- sentations integrate the current field of view with the remembered sition. The Journal of Neuroscience 28(34): 8387–8396. 360° panorama in scene-selective cortex. Current Biology 26(18): Miyashita T and Rockland KS (2007) GABAergic projections from the 2463–2468. hippocampus to the retrosplenial cortex in the rat. The European Robinson S, Todd TP, Pasternak AR, et al. (2014) Chemogenetic silenc- Journal of Neuroscience 26(5): 1193–1204. ing of neurons in retrosplenial cortex disrupts sensory precondition- Morris, RG (2006) Elements of a neurobiological theory of hippocampal ing. The Journal of Neuroscience 34(33): 10982–10988. function: the role of synaptic plasticity, synaptic tagging and sche- Ruggiero G, Frassinetti F, Iavarone A, et al. (2014) The lost ability to mas. Eur J Neurosci. 23(11):2829-46. find the way: Topographical disorientation after a left brain lesion. Mullally SL, Hassabis D and Maguire EA (2012) Scene construction Neuropsychology 28(1): 147–160. in amnesia: An fMRI study. The Journal of Neuroscience 32(16): Sestieri C, Corbetta M, Romani GL, et al. (2011) Episodic memory 5646–5653. retrieval, parietal cortex, and the default mode network: functional Murray EA and Wise SP (2010) What, if anything, can monkeys tell us and topographic analyses. The Journal of Neuroscience 31(12): about human amnesia when they can’t say anything at all? Neuropsy- 4407–4420. chologia 48(8): 2385–2405. Shine JP, Valdés-Herrera JP, Hegarty M, et al. (2016) The human retro- Neave N, Lloyd S, Sahgal A, et al. (1994) Lack of effect of lesions in splenial cortex and thalamus code head direction in a global refer- the anterior cingulate cortex and retrosplenial cortex on certain tests ence frame. The Journal of Neuroscience 36(24): 6371–6381. of spatial memory in the rat. Behavioural Brain Research 65(1): Shinkai M, Yokofujita J, Oda S, et al. (2005) Dual axonal terminations 89–101. from the retrosplenial and visual association cortices in the laterodorsal Nelson AJD, Hindley EL, Pearce JM, et al. (2015a) The effect of retro- thalamic nucleus of the rat. Anatomy and Embryology 210(4): 317– splenial cortex lesions in rats on incidental and active spatial learn- 326. ing. Frontiers in Behavioral Neuroscience 9: 11. DOI: 10.3389/ Silson EH, Steel AD and Baker CI (2016) Scene-selectivity and retino- fnbeh.2015.00011. eCollection 2015. topy in medial parietal cortex. Frontiers in Human Neuroscience 10: Nelson AJD, Powell AL, Holmes JD, et al. (2015b) What does 412. DOI: 10.3389/fnhum.2016.00412. spatial alternation tell us about retrosplenial cortex function? Smith DM, Barredo J and Mizumori SJY (2012) Complimentary roles Frontiers in Behavioral Neuroscience 9: 126. DOI: 10.3389/ of the hippocampus and retrosplenial cortex in behavioral context fnbeh.2015.00126. discrimination. Hippocampus 22(5): 1121–1133. Nestor PJ, Fryer TD, Ikeda M, et al. (2003) Retrosplenial cortex (BA Smith DM, Freeman JH Jr, Nicholson D, et al. (2002) Limbic thalamic 29/30) hypometabolism in mild cognitive impairment (prodromal lesions, appetitively motivated discrimination learning, and training- Alzheimer’s disease). The European Journal of Neuroscience 18(9): induced neuronal activity in rabbits. The Journal of Neuroscience 2663–2667. 22(18): 8212–8221. Nonaka M, Fitzpatrick R, Lapira J, et al. (2017) Everyday memory: Spiers HJ and Maguire EA (2006) Thoughts, behaviour, and brain Towards a translationally effective method of modelling the encod- dynamics during navigation in the real world. NeuroImage 31(4): ing, forgetting and enhancement of memory. The European Journal 1826–1840. of Neuroscience 46(4): 1937–1953. Spreng RN, Mar RA and Kim ASN (2009) The common neural basis of Parron C and Save E (2004) Comparison of the effects of entorhinal and autobiographical memory, prospection, navigation, theory of mind, retrosplenial cortical lesions on habituation, reaction to spatial and and the default mode: A quantitative meta-analysis. Journal of Cog- non-spatial changes during object exploration in the rat. Neurobiol- nitive Neuroscience 21(3): 489–510. ogy of Learning and Memory 82(1): 1–11. Sripanidkulchai K and Wyss JM (1986) Thalamic projections to retro- Passarelli L, Rosa MGP, Bakola S, et al. (2017) Uniformity and diversity splenial cortex in the rat. The Journal of Comparative Neurology of cortical projections to precuneate areas in the macaque monkey: 254(2): 143–165. What defines area PGm? Cerebral Cortex. Epub ahead of print 25 Sutherland RJ, Whishaw IQ and Kolb B (1988) Contributions of cingu- March. DOI: 10.1093/cercor/bhx067. late cortex to two forms of spatial learning and memory. The Journal Patai EZ, Javadi AH, Ozubko JD, et al. (2017) Long-term consolidation of Neuroscience 8(6): 1863–1872. switches goal proximity coding from hippocampus to retrosplenial Suzuki WA and Amaral DG (1994) Perirhinal and parahippocampal cortex. Available at: http://www.biorxiv.org/content/early/2017/07/ cortices of the macaque monkey: Cortical afferents. The Journal of 25/167882.1.abstract Comparative Neurology 350(4): 497–533. Mitchell et al. 13 Tamir DI and Mitchell JP (2011) The default network distinguishes con- Vann SD and Aggleton JP (2005) Selective dysgranular retrosplenial cor- struals of proximal versus distal events. Journal of Cognitive Neuro- tex lesions in rats disrupt allocentric performance of the radial-arm science 23(10): 2945–2955. maze task. Behavioral Neuroscience 119(6): 1682–1686. Taube JS (1995) Head direction cells recorded in the anterior thalamic Vann SD, Brown MW, Erichsen JT, et al. (2000) Fos imaging reveals nuclei of freely moving rats. The Journal of Neuroscience 15(1 Pt differential patterns of hippocampal and parahippocampal subfield 1): 70–86. activation in rats in response to different spatial memory tests. The Taube JS (2007) The head direction signal: Origins and sensory-motor Journal of Neuroscience 20(7): 2711–2718. integration. Annual Review of Neuroscience 30: 181–207. Vedder LC, Miller AMP, Harrison MB, et al. (2017) Retrosplenial Teipel S, Grothe MJ and Alzheimer’s Disease Neuroimaging Initiative cortical neurons encode navigational cues, trajectories and reward (2016) Does posterior cingulate hypometabolism result from discon- locations during goal directed navigation. Cerebral Cortex 27(7): nection or local pathology across preclinical and clinical stages of 3713–3723. Alzheimer’s disease? European Journal of Nuclear Medicine and Vertes RP, Albo Z, Viana Di and Prisco G (2001) Theta-rhythmically Molecular Imaging 43(3): 526–536. firing neurons in the anterior thalamus: implications for mnemonic Todd TP, Meyer HC and Bucci DJ (2015) Contribution of the retrosple- functions of Papez’s circuit. Neuroscience 104(3): 619–625. nial cortex to temporal discrimination learning. Hippocampus 25(2): Vogt BA (1976) Retrosplenial cortex in the rhesus monkey: A cytoar- 137–141. chitectonic and Golgi study. The Journal of Comparative Neurology Tsanov M, Chah E, Vann SD, et al. (2011) Theta-modulated head direc- 169(1): 63–97. tion cells in the rat anterior thalamus. The Journal of Neuroscience Vogt BA and Miller MW (1983) Cortical connections between rat cingu- 31(26): 9489–9502. late cortex and visual, motor, and postsubicular cortices. The Journal Tse D, Takeuchi T, Kakeyama M, et al. (2011) Schema-dependent gene of Comparative Neurology 216(2): 192–210. activation and memory encoding in neocortex. Science 333(6044): Vogt BA, Pandya DN and Rosene DL (1987) Cingulate cortex of the rhe- 891–895. sus monkey: I. Cytoarchitecture and thalamic afferents. The Journal Van Groen T and Wyss MJ (1990) Connections of the retrosplenial of Comparative Neurology 262(2): 256–270. granular a cortex in the rat. The Journal of Comparative Neurology Vogt BA, Vogt L and Laureys S (2006) Cytology and functionally corre- 300(4): 593–606. lated circuits of human posterior cingulate areas. NeuroImage 29(2): Van Groen T and Wyss MJ (1992a) Connections of the retrosplenial dys- 452–466. granular cortex in the rat. The Journal of Comparative Neurology Wesierska M, Adamska I and Malinowska M (2009) Retrosplenial cor- 315(2): 200–216. tex lesion affected segregation of spatial information in place avoid- Van Groen T and Wyss MJ (1992b) Projections from the laterodorsal ance task in the rat. Neurobiology of Learning and Memory 91(1): nucleus of the thalamus to the limbic and visual cortices in the rat. 41–49. The Journal of Comparative Neurology 324(3): 427–448. Whishaw IQ, Maaswinkel H, Gonzalez CL, et al. (2001) Deficits in Van Groen T and Wyss MJ (2003) Connections of the retrosplenial allothetic and idiothetic spatial behavior in rats with posterior granular B cortex in the rat. The Journal of Comparative Neurology cingulate cortex lesions. Behavioural Brain Research 118(1): 463(3): 249–263. 67–76. Van Hoesen G and Pandya DN (1975) Some connections of the entorhi- Wolbers T and Büchel C (2005) Dissociable retrosplenial and hippocam- nal (area 28) and perirhinal (area 35) cortices of the rhesus monkey, pal contributions to successful formation of survey representations. I, temporal lobe afferents. Brain Research 95(1): 1–24. The Journal of Neuroscience 25(13): 3333–3340. Vann SD, Aggleton JP and Maguire EA (2009) What does the retrosple- Wood ER, Dudchenko PA, Robitsek RJ, et al. (2000) Hippocampal neu- nial cortex do? Nature Reviews Neuroscience 10(11): 792–802. rons encode information about different types of memory episodes Vann SD and Aggleton JP (2002) Extensive cytotoxic lesions of the rat occurring in the same location. Neuron 27(3): 623–633. retrosplenial cortex reveal consistent deficits on tasks that tax allo- Wyss J and Van Groen T (1992) Connections between the retrosplenial centric spatial memory. Behavioral Neuroscience 116(1): 85–94. cortex and the hippocampal formation in the rat: A review. Hippo- Vann SD and Aggleton JP (2004) Testing the importance of the retrosple- campus 2(1): 1–11. nial guidance system: Effects of different sized retrosplenial cortex Yamawaki N, Radulovic J and Shepherd GMG (2016) A corticocortical lesions on heading direction and spatial working memory. Behav- circuit directly links retrosplenial cortex to M2 in the mouse. The ioural Brain Research 155(1): 97–108. Journal of Neuroscience 36(36): 9365–9374.

Journal

Brain and Neuroscience AdvancesSAGE

Published: Mar 19, 2018

Keywords: Learning; memory; cingulate cortex; primate; hippocampal formation; thalamus; neuroimaging; default mode network; immediate-early genes; electrophysiology

There are no references for this article.