Differential functions of the dorsal and intermediate regions of the hippocampus for optimal goal-directed navigation in VR space

Hyeri Hwang; Seung-Woo Jin; Inah Lee

doi:10.7554/eLife.97114

. 2024 Jul 16;13:RP97114. doi: 10.7554/eLife.97114

Differential functions of the dorsal and intermediate regions of the hippocampus for optimal goal-directed navigation in VR space

Hyeri Hwang ¹, Seung-Woo Jin ², Inah Lee ^1,^✉

Editors: Laura L Colgin³, Laura L Colgin⁴

PMCID: PMC11251721 PMID: 39012807

Abstract

Goal-directed navigation requires the hippocampus to process spatial information in a value-dependent manner, but its underlying mechanism needs to be better understood. Here, we investigated whether the dorsal (dHP) and intermediate (iHP) regions of the hippocampus differentially function in processing place and its associated value information. Rats were trained in a place-preference task involving reward zones with different values in a visually rich virtual reality environment where two-dimensional navigation was possible. Rats learned to use distal visual scenes effectively to navigate to the reward zone associated with a higher reward. Inactivation of both dHP and iHP with muscimol altered the efficiency and precision of wayfinding behavior, but iHP inactivation induced more severe damage, including impaired place preference. Our findings suggest that the iHP is more critical for value-dependent navigation toward higher-value goal locations.

Research organism: Rat

Introduction

It has long been suggested that the hippocampus is the neural substrate of a ‘cognitive map’ – a map-like representation of the spatial environment that allows flexible spatial navigation (O’Keefe and Nadel, 1978). The cognitive map is also needed for remembering important events in space. Animals in the natural environment often navigate to achieve goals, such as finding food or avoiding predators, and this goal-directed navigation involves remembering places and their associated values. It has been reported that the receptive fields of place cells in the hippocampus tend to accumulate near a goal location or shift toward it (Hollup et al., 2001; Kennedy and Shapiro, 2009; Dupret et al., 2010). One could argue that the hippocampus must process task- or goal-relevant information, including the value of a place, to achieve the goal. However, the specific hippocampal processes involved in integrating the two types of representation – place and value – toward goal-oriented behavior are still largely unknown.

Such integration may occur along the dorsoventral axis of the hippocampus. Previous anatomical studies (Krettek and Price, 1977; Swanson et al., 1978; Pikkarainen et al., 1999; Tao et al., 2021) suggest that the hippocampus can be divided along its dorsoventral axis into dorsal (dHP), intermediate (iHP), and ventral (vHP) hippocampal subregions based on different anatomical characteristics. The dHP is connected with brain regions that process visuospatial information, including the retrosplenial cortex and the caudomedial entorhinal cortex (Van Groen and Wyss, 2003; Dolorfo and Amaral, 1998); it also communicates with the iHP via bidirectional extrinsic connections but exhibits limited connections with the vHP (Tao et al., 2021; Swanson et al., 1978). The iHP receives heavy projections from valence-related areas, such as the amygdala and ventral tegmental area (VTA) – subcortical inputs that are less prominent in the dHP (Pikkarainen et al., 1999; Felix-Ortiz and Tye, 2014; Gasbarri et al., 1994). The vHP also has connections with the iHP and value-representing areas such as the amygdala, but it does not project heavily to the dHP (Tao et al., 2021; Swanson et al., 1978; Pikkarainen et al., 1999; Krettek and Price, 1977). Notably, compared with the dHP, the iHP and vHP have much heavier connections with the medial prefrontal cortex (mPFC), contributing to goal-directed action control (Hoover and Vertes, 2007; Liu and Carter, 2018). Additionally, the three subregions along the dorsoventral axis display different gene expression patterns that corroborate the anatomical delineations (Dong et al., 2009; Bienkowski et al., 2018). Overall, the iHP subregion of the hippocampus appears to be ideally suited to integrating information from the dHP and vHP.

Surprisingly, beyond the recognition of anatomical divisions, the available literature on the functional differentiation of subregions along the dorsoventral axis of the hippocampus, particularly in the context of value representation, is somewhat inconsistent. Specifically, there is physiological evidence that the size of a place field becomes larger as recordings of place cells move from the dHP to the vHP (Jung et al., 1994; Maurer et al., 2005; Kjelstrup et al., 2008; Royer et al., 2010). Thus, it has been thought that the dHP is more specialized for fine-grained spatial representation than the iHP and vHP. However, when it comes to the neural representation of value information of a place, results are mixed. Several studies have reported that place fields recorded in the dHP respond to internal states and motivational significance based on their accumulation near behaviorally significant locations (e.g., reward locations or escape platforms; Hollup et al., 2001; Kennedy and Shapiro, 2009; Dupret et al., 2010), or to the reward per se (Gauthier and Tank, 2018). In contrast, others have reported that dHP place cells do not alter their activity according to a change in reward or reward location and thus do not represent value information (Duvelle et al., 2019; Jin and Lee, 2021; Speakman and O’Keefe, 1990).

Furthermore, although the iHP and vHP have mainly been studied in the context of fear and anxiety, several studies have also reported spatial representation and value-related signals in these subregions. Specifically, prior studies reported that rats with a dysfunctional dHP retained normal goal-directed, target-searching behavior if the iHP and vHP were intact (Moser et al., 1995; de Hoz et al., 2003). Moreover, lesions in the iHP have been shown to impair rapid place learning in the water maze task (Bast et al., 2009). Our laboratory also reported that place cells in the iHP, but not the dHP, instantly respond to a change in spatial value and overrepresent high-value locations (Jin and Lee, 2021).

Based on existing experimental evidence, we hypothesize that the iHP is the primary locus for associating spatial representation with value information, distinguishing it from the dHP and vHP. In the current study, we investigated the differential functions of the dHP and iHP in goal-directed spatial navigation by monitoring behavioral changes after pharmacological inactivation of either of the two regions as rats performed a place-preference task in a two-dimensional (2D) virtual reality (VR) environment. In this experimental paradigm, rats learned to navigate toward one of two hidden goal locations associated with different reward amounts. Whereas inactivation of the dHP mainly affected the precision of wayfinding, iHP inactivation impaired value-dependent navigation more severely by affecting place preference.

Results

Well-trained rats align themselves toward the high-value zone before departure in the place-preference task

We established a VR version of a place-preference task (Figure 1A) in which rats could navigate a 2D environment by rolling a spherical treadmill with their body locations fixed, allowing them to run at the apex of the treadmill. Body-fixed rats (n = 8) were trained to explore a virtual circular arena surrounded by multiple distal visual landmarks (houses, rocks, mountains, and trees) (Figure 1B). Rats were always started at the center of the arena, and the arena contained two unmarked reward zones – a high-value zone and a low-value zone – each associated with different amounts of honey water (6:1 ratio between high- and low-value zones). A trial started with the rat facing one of six start directions – north (N), northeast (NE), southeast (SE), south (S), southwest (SW), and northwest (NW) – determined pseudorandomly to guarantee equal numbers of trials in all directions. In the example trial shown in Figure 1C, the rat was heading in the NE direction at the start location (‘Trial Start’), then turned to the left side to run toward the W goal zone (‘Navigation’). Once the rat arrived at one of the reward zones, the synchronization between the spherical treadmill movement and the virtual environment stopped, and multiple drops of honey water were delivered via the licking port (‘Reward’). Then, during an inter-trial interval (ITI), the LCD screens turned gray, and the rat was required to remain still for 5 s to initiate the subsequent trial. The pre-surgical training session consisted of 60 trials, which were reduced to 40 during post-surgical training.

Pre-surgical training began after shaping in the VR environment. On average, it took 13 days for rats to reach pre-surgical criteria, namely, to complete 60 trials and visit the high-value goal zone in more than 75% of completed trials (see ‘Materials and methods’ for the detailed performance criteria). Well-trained rats exhibited two common behaviors during the pre-surgical training. First, although it was not required in the task, they learned to rotate the spherical treadmill counterclockwise to move around in the virtual environment (presumably to perform energy-efficient navigation). To rule out the potential effect of hardware bias or any particular aspect of peripheral landscape to make rats turn only to one side, we measured the direction of the first body-turn in each trial on the last day of shaping and the first day of the main task (i.e., before rats learned the reward zones). There was no significant difference between the clockwise and counterclockwise turns (p=0.46 for shaping, p=0.76 for the main task; Wilcoxon signed-rank test), indicating that the stereotypical pattern of counterclockwise body-turn appeared only after the rats learned the reward locations.

Second, once a trial started, the animal rotated the treadmill immediately to align its starting direction with the visual scene associated with the high-value reward zone. After setting the starting direction, the rat started to run on the spherical treadmill, moving the treadmill forward to navigate directly toward the reward zone. All eight rats displayed this strategy during the later learning phase but not during the earlier learning stage, suggesting that the start-scene–alignment strategy was learned during training. Because the initial rotational scene alignment before departure was an essential component of the task and this behavior was not readily detectable with position-based analysis, we based most of our behavioral analysis on the directional information defined by the allocentric reference frame of the virtual environment (Figure 2A). Because we did not measure the rat’s head direction in the current study, the allocentric directional information represented the angular position of a particular scene in the virtual environment displayed in the center of the screen.

Figure 2. — (A) The reference frame of the virtual environment. The six start directions are illustrated with the red high-value zone (180°) and blue low-value zone (0°). On the right, the departure circle (DC) is denoted with a purple dashed line, and the start direction is marked with a black arrowhead and a green arrow. (B) Overall changes in scene direction over the normalized distance between the start location and the DC (left). Each colored line indicates the median change of scene direction in trials with each start location, and red and blue arrowheads mark high- and low-value zone centers, respectively. The 0°-to-360° range was repeated in the ordinate of the plot to capture rotational movements in opposite directions (positive and negative directions for clockwise and counterclockwise rotations, respectively). The gray lines on the right show the rat’s trajectory within the DC. These examples were excerpted from the first and last days of pre-training of a single rat. The numbers after ‘Novice’ and ‘Expert’ indicate the rat and session number of the example. (C) Individual examples of scene directions and trajectories in the novice session. Scene direction change for each direction is drawn separately (top) for individual trials. The black arrowhead indicates that specific start direction. Trajectories within the DC (middle) and the whole arena (bottom) are also illustrated according to the indicated color code. Mean travel distance in meters and latency in seconds are shown below the virtual reality (VR) arena trajectory. (D) Same as (C), but for the expert session.

To establish the direction in which the rat departed the starting point at the center after scene alignment, we first defined a departure circle – a virtual circle (~20cm in diameter) in the VR environment at the center of the arena (Figure 2A). In the example trial shown in Figure 2A, the rat faced the NE direction (315°) at the trial start but immediately turned its body to the NW direction upon starting and ran straight toward the high-value zone after that. Since the initial scene rotations at the start point cannot be visualized in the position-based graph, we made a scene rotation plot that visualizes the rotational movement traces in the virtual environment. The scene rotation plot covers the period from the start of the trial to when the rat leaves the departure circle (Figure 2B).

On the first day of training for the task (‘Novice’ stage), the rat produced almost no rotation of the VR environment until he exited the departure circle, indicating that the animal ran straight in the initially set start direction without adjusting the scene orientation. As a result, rats missed the target reward zones in most trials (Figure 2B and C). However, by the last day of training (‘Expert’ stage), there were noticeable rotational shifts in all directional traces (i.e., counterclockwise rotations) that converged on the high-value reward zone (Figure 2B and D). This was the case for all trials except those in which the initial start direction almost matched the orientation of the high-value reward zone (i.e., 225° or NW). Furthermore, the average travel distance and latency for each start direction declined from the novice to the expert stage, suggesting that the rats navigated more efficiently toward reward zones in the later stage of learning by pre-adjusting their starting scene direction at the trial start (Figure 2C and D).

Overall, the marked differences in orienting behaviors between early and late learning stages suggest that rats could discriminate the high-value reward zone from the low-value zone in our VR environment and show that they preferred visiting the high-value reward zone over the low-value zone. It also indicates that rats could explore the VR environment using allocentric visual cues to find the critical scenes associated with the high-value zone before leaving the starting point (i.e., departure circle).

Departing orientation and perimeter-crossing direction provide a measure of navigational efficiency and precision, respectively

To analyze behavioral changes during learning in more detail, we analyzed various learning-related parameters at different stages of pre-surgical training. For this, we focused on days in which rats visited the high-value zone on more than 75% of trials for two consecutive days – the performance criterion for completion of pre-surgical training. These two consecutive days (post-learning days) were grouped and averaged for each rat as the post-learning group (‘POST’ in Figure 3Aii) and compared with the two consecutive days immediately preceding the post-learning days (pre-learning days; ‘PRE’ in Figure 3Aii).

We first measured the departing direction when crossing the departure circle (departing direction [DD]; Figure 3Ai). As indicated in Figure 2, well-trained rats rotated the VR environment to place the target VR scene (i.e., high-value reward zone scene) ahead before departure. Therefore, alignment of the DD with the high-value zone at the beginning of navigation indicated that the rat remembered the scenes associated with the high-value zone. For example, the distribution of pre-learning days DDs was widely distributed without any directional bias; as such, its mean vector was small (Figure 3Aii). On the other hand, the DDs of post-learning sessions mostly converged on the direction aligned with the high-value zone, resulting in a larger mean vector length compared with that in the pre-learning session. The distributions of averaged DDs for all rats significantly differed between ‘PRE’ and ‘POST’ (p<0.001, Kuiper’s test), verifying that DD is a valid index of the acquisition of the high-value zone and efficient navigation. To investigate how accurately rats oriented themselves directly to the high-value zone before leaving the start point, we also calculated the average deviation angle of DDs (DD-deviation) – the angle between the DD and the high-value zone (180°, measured at the center of the zone) – for each rat (Figure 3Aiii). A comparison between pre- and post-learning sessions showed that the DD-deviation significantly declined after learning. This implies that well-trained rats aligned their bodies more efficiently to directly navigate to the high-value zone (p<0.01, Wilcoxon signed-rank test).

Rats adjusted their navigational routes further, even after exiting the departure circle, to navigate more accurately and straight to the goal, avoiding the wall surrounding the arena. Such fine spatial tuning (i.e., navigation precision), measured as the decrease in DD-deviation, only appeared after the rats learned the high-value reward location. To quantify navigation precision, we measured the perimeter-crossing direction (PCD; Figure 3Bi), defined as the angle at which the rat first touched the unmarked circular boundary along the arena’s perimeter, which shares the inner boundaries of the reward zones (green dashed lines in Figure 3Bi). During pre-learning, the PCD was randomly distributed along the perimeter (‘PRE’ in Figure 3Bii). On the other hand, in most post-learning stage trials, rats crossed the unmarked peripheral boundaries only in the vicinity of the high-value zone (‘POST’ in Figure 3Bii). Since rats usually turned counterclockwise during navigation, the convergence of crossings near the northern edge of the high-value zone indicates that they took a shortcut – the most efficient route – to enter the goal zone. The PCD distributions were significantly different between pre- and post-learning stages (p<0.001, Kuiper’s test) (Figure 3Bii). The deviation angle between the PCD and the high-value zone center also significantly decreased with learning (Figure 3Biii), indicating that the navigation of rats to the goal became more accurate.

Additionally, to investigate whether the rats used a certain landmark as a beacon to find the reward zones, we conducted the landmark omission test as a part of control experiments. Here, one of the landmarks was omitted, and the landmark to be made disappear was pseudorandomly manipulated on a trial-by-trial basis. The omission of one landmark, regardless of its identity, did not cause a specific behavioral change in finding the reward zones, suggesting that the rats were not relying on a single visual landmark when finding the reward zones. The result can be reported anecdotally only because of an insufficient sample size (n = 3), not permitting any meaningful statistical testing.

Navigation is impaired by inactivation of either the dHP or iHP, but only iHP inactivation affects place-preference behavior

To dissociate the roles of the dHP and iHP, we inactivated either the dHP or iHP in an individual animal using muscimol (MUS), a GABA-A receptor agonist, before the rat performed the place-preference task. To allow within-subject comparisons in performance, we bilaterally implanted two pairs of cannulas – one targeting the dHP and the other targeting the iHP – in the same rat after it successfully reached pre-surgical training criteria (Figure 4A). After 1 week of recovery from surgery, rats were retrained to regain a level of performance similar to that in the pre-surgical training period (Figure 4B), after which the drug injection schedule was started.

Figure 4. — (A) Cannula positions marked. The scale bar at the upper left indicates 1mm. (i) Example of bilaterally implanted cannula tracks in Nissl-stained sections in the dorsal hippocampal (dHP) and intermediate hippocampal (iHP). (ii) Tip locations illustrated in the atlas, with different colors for individual rats (n = 8). (B) Training schedule. Rats were divided into two groups (n = 4/group) to counterbalance the injection order for the main task and probe test.

We divided rats into two drug injection groups (n = 4 rats/group) to counterbalance the injection order between the dHP and iHP. Rats in one group received drug infusion into the dHP first, whereas rats in the other group were injected into the iHP first. For all rats, phosphate-buffered saline (PBS) was initially injected in both regions as a vehicle control. For analytical purposes, we first ensured no statistical difference in performance between the two PBS sessions (dPBS and iPBS; see below) and then averaged them into a single PBS session to increase statistical power. During the PBS session, rats tended to take the most efficient path to the high-value zone, as they had done during pre-surgical training (Figure 5A). They aligned the VR scene at the start with the high-value zone for all start directions and then ran directly toward the goal zone. Notably, once the start scene alignment was complete, rats usually moved quickly and straight without slowing in the middle. Also, their navigation paths led them directly toward the center of the goal zone. During subsequent dHP-inactivation sessions, rats appeared less accurate, bumping into the arena wall in many trials (dMUS in Figure 5A), but most of these wall bumps occurred in the vicinity of the high-value zone, and rats quickly compensated for their error by turning their bodies to target the reward zone correctly after wall bumping. In contrast, in iHP-inactivation sessions, the trajectories were largely disorganized, and the wall-bumping locations were no longer limited to the vicinity of the high-value zone. In some trials, rats moved largely randomly (as shown in 860-17-24 in Figure 5A) and appeared to visit the low-value zone significantly more than during PBS or dMUS sessions.

To quantitatively analyze these observations, we compared the proportions of visits to the high-value zone among drug conditions (Figure 5B), finding a significant difference in the percentage of correct target visits among drug conditions (F_(2,14) = 10.56, p<0.01, one-way repeated-measures ANOVA; p=0.25 for dPBS vs. iPBS, Wilcoxon signed-rank test). A post hoc analysis revealed a significant decrease in the iMUS session compared to the PBS session (p<0.05, Bonferroni-corrected post hoc test). In contrast, no significant differences were found in other conditions, although there was a decreasing trend in the iMUS compared to PBS (p=0.2 for PBS vs. dMUS; p=0.1 for dMUS vs. iMUS). These results indicate that dHP-inactivated rats showed a strong preference for the high-value zone, as they did in control sessions, but that the performance of iHP-inactivated rats was impaired in our place-preference task, as reflected in their significantly more frequent visits to the low-value zone compared with controls.

It is unlikely that these differences stemmed from generic sensorimotor impairment as a result of MUS infusion because running speed remained unchanged across drug conditions (F_(2,14) = 0.99, p=0.37, one-way repeated-measures ANOVA; p=0.95 for dPBS vs. iPBS, Wilcoxon signed-rank test) (Figure 5C). Furthermore, rats remained motivated throughout the testing session, as evidenced by the absence of a significant difference in the number of trials across drug groups (F_{(1.16, 8.13)}=1.34, p=0.29, one-way repeated-measures ANOVA with Greenhouse–Geisser correction; p=1.0 for dPBS vs. iPBS, Wilcoxon signed-rank test; data not shown), although there was an increase in the session duration (F_(2,14) = 6.46, p<0.05, one-way repeated-measures ANOVA; p=0.27 for dPBS vs. iPBS, Wilcoxon signed-rank test; data not shown) during MUS sessions (dMUS and iMUS) compared with the PBS session (p-values <0.05, Bonferroni-corrected post hoc test). This increase in session duration was attributable to arena wall bumping events, which usually entailed a recovery period before rats left the peripheral boundaries and moved again toward the goal. These observations indicate that inactivation of the iHP significantly impairs the rat’s ability to effectively navigate to the higher-value reward zone in a VR environment without affecting goal-directedness or locomotor activity.

To determine how effectively rats traveled to the goal in each condition, we also quantified the errors made in each condition by assessing the number of perimeter crossings (Figure 5D). To avoid duplicate assessments, we only counted an event as a perimeter crossing when the rat crossed the perimeter boundary from inside to outside. Rats tended to make more errors in dMUS sessions compared with controls, and errors were even more prevalent in iMUS sessions (F_(2,14) = 18.59, p<0.001, one-way repeated-measures ANOVA; p=0.39 for dPBS vs. iPBS, Wilcoxon signed-rank test; p<0.01 for PBS vs. dMUS, p<0.01 in PBS vs. iMUS; p<0.05 for dMUS vs. iMUS, Bonferroni-corrected post hoc test). During PBS sessions, navigation was mostly precise, resulting in just one perimeter crossing. In the dMUS sessions, precision declined, but the rats were relatively successful in finding the high-value zone, with most trials being associated with a slightly increased number of perimeter crossings. In contrast, rats in the iMUS sessions failed to find the high-value zone. They seemed undirected, exhibiting a significantly increased number of perimeter crossings compared with the other two sessions. Taken together, these results indicate that iHP inactivation more severely damages normal goal-directed navigational patterns than dHP inactivation in our place-preference task.

The iHP causes more damage to value-dependent spatial navigation than the dHP, which is important for navigational precision

To further differentiate among conditions, we examined DD and PCD – indices of the effectiveness and precision of navigation, respectively (Figure 3). We first investigated the distribution of DDs in all trials for all rats and calculated the resultant mean vector (Figure 6A). Note that dPBS and iPBS sessions were separately illustrated here for better visualization of changes in behavioral pattern for each subregion. Whereas DDs for both PBS sessions (dPBS and iPBS) were distributed relatively narrowly toward the high-value zone, those for dMUS sessions were more widely distributed, and their peak pointed away from the reward zone. In the case of the iHP-inactivation session, some DDs were even pointed toward the opposite side of the target goal zone (i.e., the low-value zone). Thus, the mean vectors from PBS sessions were relatively longer than those from MUS sessions. The mean vectors for PBS sessions also stayed within the range of the high-value zone, whereas those for MUS sessions pointed either toward the edge of the reward zone (dMUS) or the outside of the reward zone (iMUS).

Figure 6. — (A) Grouped comparison of departing direction (DD) in each drug condition. Distributions of DDs in each drug condition (rose plots) and a comparison of their mean directions. Gray plots, PBS sessions; green plots, dHP inactivation; orange plots, iHP inactivation. Red and blue arcs indicate high- and low-value zones, respectively. Statistically significant differences in mean vectors, illustrated as arrows, are indicated with asterisks. The mean directions of all four conditions were first compared together (F_(3,1253) = 7.78, p<0.001, Watson–Williams test); a post hoc pairwise comparison was subsequently applied if the average mean vector length of the two sessions was greater than 0.45 (p<0.05 for dPBS vs. dMUS; p<0.001 for iPBS vs. iMUS; p=0.66 for dPBS vs. iPBS; Watson–Williams test). The number on the upper-right side of the plot shows the length of the mean vector. (**B, C**) Changes in mean vector length (F_(2,14) = 12.64, p<0.001, one-way repeated-measures ANOVA; p=0.24 for PBS vs. dMUS, p<0.01 for PBS vs. iMUS, p<0.05 for dMUS vs. iMUS, Bonferroni-corrected post hoc test) (B) and deviation angles from the high-value zone center (F_(2,14) = 13.37, p<0.001, one-way repeated-measures ANOVA; p=0.19 for PBS vs. dMUS, p<0.01 for PBS vs. iMUS, p<0.05 for dMUS vs. iMUS) (C) of the DD in each drug session. Error bars indicate SEMs (n=8), and the significance level was set at α = 0.05. *p<0.05, **p<0.01, ***p<0.001.

We next quantitatively confirmed these observations, comparing the mean direction for each drug condition to determine how inactivation affected the accuracy of the body alignment of rats at departure (Figure 6A). A Watson–Williams test indicated that the mean angles of DDs in all four drug conditions (dPBS, iPBS, dMUS, and iMUS) for all rats significantly differed from each other (F_(3,1253) = 7.78, p<0.001). Post hoc pairwise comparisons showed that inactivation of either the dHP or iHP significantly altered DDs compared with the PBS condition (p<0.05 for dPBS vs. dMUS; p<0.001 for iPBS vs. iMUS; p=0.66 for dPBS vs. iPBS; Watson–Williams test). Moreover, the mean DDs of dMUS and iMUS sessions were displaced from the center of the high-value zone compared with those of PBS sessions (i.e., dPBS and iPBS), suggesting that the rats did not accurately align themselves to the target reward zone at the time of departure. The mean vector of the iMUS session also appeared smaller than that of the other conditions, indicating a less concentrated distribution of DDs with iHP inactivation. Unfortunately, it was not possible to perform a statistical comparison between dMUS and iMUS because the DDs for the iMUS session were too dispersed to yield a mean vector with a sufficient length to compare directions between the two conditions (averaged mean vector length of dMUS and iMUS sessions <0.45; Berens, 2009).

The mean vector lengths for DDs were also significantly different among drug conditions (F_(2,14) = 12.64, p<0.001, one-way repeated-measures ANOVA; p=0.55 for dPBS vs. iPBS, Wilcoxon signed-rank test) (Figure 6B), with a post hoc analysis showing a significant difference between the iMUS session and both PBS (p<0.01) and dMUS (p=0.05) sessions; however, no significant difference was found between PBS and dMUS sessions (p=0.24, Bonferroni-corrected post hoc test). The profound performance deficits in the iHP-inactivated condition were also confirmed by examining the DD-deviation from the target direction, defined as the center of the high-value zone (Figure 6C). Specifically, we found that DD-deviations were significantly different among drug conditions (F_(2,14) = 13.37, p<0.001, one-way repeated-measures ANOVA; p=0.38 for dPBS vs. iPBS, Wilcoxon signed-rank test), with a Bonferroni-corrected post-hoc test revealing a significant increase in DD-deviation in the iMUS session compared with both the PBS session (p<0.01) and the dMUS session (p<0.05). Again, no significant difference was found between PBS and dMUS sessions (p=0.19). These results demonstrate that disruption of the dHP does not significantly affect the ability of rats to orient themselves effectively at departure to target the high-value reward zone. In contrast, inactivation of the iHP across all trials caused rats to depart the starting location without strategically aligning to the scene and consequently failing to hit the target zone effectively and directly.

Next, we ran similar analyses for the PCD (Figure 7), also investigating the PCD distribution and its mean vector for each drug condition (Figure 7A). PCD distributions appeared similar to those for DD; the PCD distributions of PBS sessions were narrowly contained within the high-value reward zone, whereas those of MUS sessions were more dispersed and misaligned with the reward zone. Again, the PCD distribution of the iMUS session showed some occurrences near the low-value zone. An examination of the resulting mean vectors using a Watson–Williams test revealed a significant difference in mean PCD angle in all sessions except for the comparison between the two PBS sessions (F_(3,1253) = 16.22, p<0.001; p=0.08 for dPBS vs. iPBS). The mean PCD angle of the dMUS session was shifted toward the upper end of the high-value zone (p<0.001 for dPBS vs. dMUS), whereas that of the iMUS session was outside of the reward zone (p<0.001 for iPBS vs. iMUS). Notably, iHP inactivation resulted in more severe errors in finding the high-value zone than dHP inactivation (p<0.01 for dMUS vs. iMUS). Interestingly, with iHP inactivation, several PCDs were found near the low-value zone, an outcome that rarely occurred in other conditions. Considering the decreased percentage of high-value zone visits (Figure 5), some of these trials ended with the rat visiting the low-value zone, suggesting an impaired ability of the animal to perform goal-directed navigation strategically.

The PCD mean vector length was largest in the PBS condition, shortest in the iMUS condition, and intermediate in the dMUS condition (F_(2,14) = 15.67, p<0.001, one-way repeated-measures ANOVA; p=0.55 for dPBS vs. iPBS, Wilcoxon signed-rank test) (Figure 7B). Unlike the mean vector length for DD, the PCD mean vector length differed between PBS and dMUS sessions, suggesting that wayfinding behavior was explicitly disrupted by dHP inactivation, albeit to a lesser extent compared with iHP inactivation (p<0.05 for PBS vs. dMUS; p<0.01 for PBS vs. iMUS; p=0.06 for dMUS vs. iMUS, Bonferroni-corrected post hoc test).

On the other hand, the PCD deviation angle from the center of the high-value zone increased in inverse order: smallest in the PBS condition and largest in the iMUS condition (F_(2,14) = 17.24, p<0.001, one-way repeated-measures ANOVA; p=0.55 for dPBS vs. iPBS, Wilcoxon signed-rank test) (Figure 7C). Similar to the PCD mean vector length data, the significant increase in deviation angle after dHP inactivation indicates that dHP-inactivated rats failed to achieve fine spatial tuning toward the high-value zone compared with controls (p<0.05 for PBS vs. dMUS, Bonferroni-corrected post hoc test). iHP inactivation also resulted in less accurate navigation, including perimeter crossings – effects that were more severe than those caused by dHP inactivation (p<0.01 for PBS vs. iMUS; p=0.06 for dMUS vs. iMUS, Bonferroni-corrected post hoc test).

Overall, results based on the PCD measure revealed that dHP-inactivated rats showed decreased precision in arriving at the goal, as reflected in the significant deviation of their PCD from the high-value zone. The PCD distribution was also not as narrow as under control conditions. Notably, deficits in navigation performance were even more severe in rats with iHP inactivation, and their performance impairment was qualitatively different from that observed with dHP inactivation in terms of both efficiency and precision of navigation. Again, these results suggest that, while the dHP is essential for accurate wayfinding, the iHP is crucial for value-dependent navigation to the higher-reward location.

Hippocampal inactivation does not impair cue-guided navigation or goal-directedness

After the drug injection stage, we trained five of the same rats used in the main task in a visual cue-guided navigation task to verify whether MUS inactivation of the hippocampus resulted in deficits in goal-directed navigation in general (Figure 8Ai). We used the same circular arena from the main task but removed all allocentric visual landmarks. The rat was started from a fixed location near the periphery of the arena. As the trial started, a spherical visual landmark with a checkered pattern flickered on either the left or right side (pseudorandomized across trials) of the rat’s starting position, serving as a beacon. When the rat arrived at the landmark area (see ‘Materials and methods’), the connection between treadmill movement and the virtual environment stopped, and honey water was provided as a reward. The rewards provided by left and right reward zones were the same in terms of both quality and quantity.

Figure 8. — (A) Object-guided navigation task as a probe test. (i) A flickering object appeared on either the left (‘Left trial’) or right (‘Right trial’) side of the screen. The start location is marked with a yellow dot, with a white arrow indicating the start direction, which remained the same for both trial types. (ii) Example of trajectories in one session. Blue and red lines represent trajectories from left and right trials that directly arrived at the reward zones, whereas gray lines indicate failed trials. Green dashed lines denote reward zones. (B) Comparison of the proportion of each drug condition’s direct hit trials (both left and right; F_(2,8) = 1.60, p=0.26, one-way repeated-measures ANOVA). Error bars indicate SEMs (n=5), and the significance level was set at α = 0.05.

In this version of the navigation task, the rat’s navigation was simply guided by the visual beacon, a type of task that the literature suggests is not hippocampal-dependent (Morris et al., 1986; Packard et al., 1989). Rats learned the task rapidly. Specifically, it took 3 days on average for rats to reach the criterion of completing 40 trials with an excess travel distance of less than 0.1 m (see ‘Materials and methods’). Moreover, examining their trajectories suggested that rats had no problem moving toward the visual landmark, whether it appeared on the left or right of the starting location (Figure 8Aii). Rats arrived at the reward zone directly in most trials (‘Direct hit’) but bumped into the arena wall in some trials. Given the presence of a strong visual landmark, which served as a beacon, trials in which the rat bumped the arena walls were considered failed trials (Figure 8Aii).

Finally, we applied the same drug injection schedule for the main task after the rats reached the abovementioned criterion. A one-way repeated-measures ANOVA revealed that the proportion of direct hit trials did not significantly differ across drug conditions, indicating no significant change in navigation precision when the goal was marked by the visual beacon (F_(2,8) = 1.60, p=0.26; p=0.50 for dPBS vs. iPBS, Wilcoxon signed-rank test) (Figure 8B). These results also imply that no generic sensorimotor or motivational deficits were involved. Collectively, these observations confirm that MUS injections in the hippocampus do not alter the ability of the rat to move around freely in the VR environment in a goal-directed fashion when the hippocampus is not necessary for the task.

Discussion

In the current study, we inactivated the dorsal or intermediate hippocampal region in rats performing a place-preference task in VR space to investigate the functional differentiation along the dorsoventral hippocampal axis during goal-directed navigation. Inactivation of the intermediate region, but not the dorsal region, of the hippocampus produced a marked reduction in the rat’s ability to conduct strategic goal-directed navigation in the virtual space without affecting goal-directedness or locomotor ability. We further examined navigational quality by measuring the precision of scene alignment upon departure and by assessing the efficiency (i.e., directness) of travel to the target goal zone (i.e., higher-value zone) without bumping into walls on the arena boundaries. We found that dHP-inactivated rats were modestly, but significantly, impaired not only in precisely targeting the goal at the time of departure but also in effectively traveling to the goal zone, compared with controls. Importantly, however, the ability of these dHP-inactivated rats to head toward the higher-value zone in the VR environment was unimpaired. In contrast, iHP-inactivated rats were severely impaired in the initial targeting of the goal zone at the time of departure and traveled somewhat aimlessly in the VR environment compared with both controls and dHP-inactivated rats. Our findings suggest that the dHP is essential for finding the most effective travel path for precise spatial navigation and that the iHP is necessary for navigating the space in a value-dependent manner to achieve goals.

Rats use allocentric visual scenes and landmarks to target the goal zone and adjust their paths accordingly during navigation in the VR environment

In the current paradigm, rats rotated the spherical treadmill counterclockwise immediately after the trial started at the VR arena’s center, presumably to find the visual scene to guide them directly toward the goal zone (i.e., high-value zone) upon departure. This initial orientation of departure – or DD – seems critical in our task, as evidenced by the fact that, during training, rats that miscalculated the DD usually bumped into the wall and had to reorient themselves at various positions within the environment. Once the rats learned the task, they oriented themselves before leaving the start point by rotating the visual environment until they found the goal-associated visual scenes and then ran straight toward the goal zone. These behavioral characteristics suggest that our task is heavily dependent on the rat’s ability to use the allocentric reference frame of the visual environment.

Prior studies suggest that, in an environment where the directional information comes largely from allocentric visual cues, the spiking activity of place cells is significantly modulated by directional visual cues, a finding that holds in both real and virtual environments (Acharya et al., 2016; Ravassard et al., 2013; Aronov and Tank, 2014). In one of these studies (Acharya et al.), directional modulation of place cells was observed even during random foraging in the absence of a goal-directed memory task. Notably, this was also true for spatial view cells in nonhuman primates (Rolls and O’Mara, 1995; Georges-François et al., 1999). Although we did not record place cells in our study, hippocampal place cells could be predicted to exhibit directional firing patterns associated with the visual scenes along the periphery of the current VR environment. Because rats in the current study rotated the environment until they found the target-matching scene without leaving the center starting point, our VR task may be an ideal behavioral paradigm for examining the directional firing of place cells in future studies.

Inactivating the dHP impairs navigational precision but does not affect place preference based on differential reward values

Our working model posits that the dHP represents a fine-scaled spatial map of an environment, in this case, a VR environment, that allows an animal to map its location precisely and choose the most efficient travel routes. Our experimental results support this model, demonstrating that dHP-inactivated rats deviated slightly, but significantly, from the ideal target heading at the time of departure (measured by DD), resulting in crossing the area boundary near the target goal zone. Nonetheless, it is important to note that dHP-inactivated rats in our study oriented themselves normally in the direction of the high-value reward zone at the time of departure, suggesting that the value-coding cognitive map and its use were intact and able to spatially guide the rats to the high-reward area in the absence of a functioning dHP. We argue that such intact place-preference performance with reasonable spatial navigation ability is supported by the iHP (presumably in connection with the vHP) in dHP-inactivated rats.

Whether the dHP represents value signals remains a matter of controversy. According to previous studies, place fields of the dHP seem to translocate to or accumulate near the location with motivational significance (e.g., reward zone), and where the strategic importance is higher (e.g., choice point in the T-maze) (Hollup et al., 2001; Lee et al., 2006; Kennedy and Shapiro, 2009; Dupret et al., 2010; Ainge et al., 2011; Valenti et al., 2018). For instance, the overrepresentation of the escape platform in a water maze – a location of high motivational significance – was observed in the neural firing patterns of place cells in the hippocampus (Hollup et al., 2001). In addition, Lee et al. reported that dHP place fields gradually translocate toward the goal arm of a continuous T-maze (Lee et al., 2006), and Dupret and colleagues suggested a goal-directed reorganization of hippocampal place fields based on an experimental paradigm in which reward locations were changed daily (Dupret et al., 2010). Such accumulation of spatial firing is not restricted to the goal location, as place fields recorded from the dHP were reported to be unevenly distributed near the start box and the choice point of a T-maze (Kim et al., 2012). One potential explanation for the discrepancy between our study and studies that reported apparent valence-dependent signals in the dHP could be that the dHP processes motivational and strategic significance (from the perspective of task demand), which is not always the same as the reward. Significance might include task demand, such as a change between random and directed search of reward (Markus et al., 1995), or a change in a significant environment stimulus from which the goal location needs to be calculated (Gothard et al., 1996). However, none of these were changed in our experimental paradigm, which might explain why dHP inactivation did not affect place-preference behavior.

Another possibility is that the dHP responds only to a more radical change in value, such as the presence or absence of reward, but not to different amounts of the same reward. Indeed, hippocampal neuronal activity does not show an explicit response to reward value in rats trained to visit arms of a plus-maze in descending order of reward amount (Tabuchi et al., 2003). Moreover, when the reward is unexpectedly altered to a less preferred one, thus decreasing motivational significance, place cells in the dHP remain mostly unchanged (Jin and Lee, 2021). These results suggest that the dHP is not crucial to maintaining value preference, a finding in line with the observed absence of an effect on place preference after dHP inactivation in our study. A recent study in which mice were trained to associate a particular odor with an appetitive outcome and distinguish it from the non-rewarded odor suggested that the dHP is responsible for stimulus identity, not saliency (Biane et al., 2023). This might be another possible interpretation of our dHP results since we used the same honey water reward for both reward zones.

The iHP may contain a value-associated cognitive map with reasonable spatial resolution for value-based navigation

iHP-inactivated rats showed poor goal-directed navigation, characterized by misalignment of their departing orientation with the goal zone and arrival points that were often far removed from the goal zone compared with the same rats under both control and dHP-inactivated conditions. Particularly, rats changed their heading directions during the navigation when they were not confident with the location of the higher reward, resulting in a less efficient route to the goal location. Rats showing this type of behavior tended to hit the perimeter of the arena first before correcting their routes. Therefore, when considered together with DD, our PCD measure could tell that the rats not hitting the goal zone directly after departure were impaired in orienting themselves to the target zone accurately from the start, not in maintaining the correct heading direction to the goal zone at the start location.

Although there is still a possibility that the levels of expression of GABA-A receptors might be different along the longitudinal axis of the hippocampus, these results support our working model that the iHP is critical for representing a value-associated cognitive map of the environment. iHP-inactivated rats, presumably unable to utilize such a value-representing map, could not strategically plan and organize their behaviors to target the high-value area in the current study. Consequently, the fine-grained spatial map present in the dHP may be of little use without the guidance of the value-associated map in the iHP, accounting for the poor navigational performance of iHP-inactivated rats. The value-associated cognitive map in the iHP may still show reasonable spatial specificity, as evidenced by the larger, but still specifically located, place fields in the iHP compared with the dHP (Jin and Lee, 2021).

The involvement of the iHP in spatial value association has been reported or implicated in several studies. For example, Bast and colleagues reported that rapid place learning is disrupted by removing the iHP and vHP, even when the dHP remains undamaged (Bast et al., 2009). On the other hand, if the iHP is spared but the dHP and vHP are removed by lesioning, rats in a water maze test quickly learn a new platform location normally. Moreover, a change in reward value induced an immediate global remapping response and a greater overrepresentation of the reward zone with a higher value in iHP neurons (Jin and Lee, 2021). Another recent study by Jarzebowski et al. focused on how hippocampal place cells change their firing patterns during the learning process for several sets of changing reward locations (Jarzebowski et al., 2022). The results from this study suggest that, in the iHP, the same place cells persistently fire across different reward locations, thus tracking the changes in reward locations.

Anatomically, the iHP is in an ideal position to represent associations between a space and its value by intrahippocampal connections from both the dHP and vHP (Tao et al., 2021; Swanson et al., 1978). Importantly, the vHP is known to receive much heavier projections from value-processing subcortical areas, such as the amygdala and VTA, compared with the upper two-thirds of the hippocampus (Krettek and Price, 1977; Swanson et al., 1978; Pikkarainen et al., 1999; Felix-Ortiz and Tye, 2014; Gasbarri et al., 1994). Thus, although the iHP also receives afferent projections from these areas, it is highly likely that the vHP plays a crucial role in the value-related representation of the iHP. Notably, both the amygdala and VTA are known to be involved in processing palatability information (Tye and Janak, 2007; Fontanini et al., 2009; Chen et al., 2020), and the amygdala has a subpopulation of neurons dedicated to encoding positive values (Kim et al., 2016; Beyeler et al., 2016). These anatomical studies support our working model of the iHP in integrating place-value information.

It is worth noting that the iHP sends direct projections to the mPFC, which is thought to be involved in behavioral control and action (Hoover and Vertes, 2007; Liu and Carter, 2018). Our experimental paradigm required rats to choose and navigate toward one of two reward zones with different values, a task structure that must demand active cognitive control, presumably by the mPFC in collaboration with the hippocampus. It is also possible that inactivation of the iHP prevents the transfer of the dHP’s spatial information to the mPFC via the iHP, which may explain why iHP inactivation produces severe deficits in goal-directed navigation in the current task. Based on these findings, we propose a working model in which the iHP associates spatial value information with the cognitive map of the dHP and sends value-associated spatial information to the mPFC, which translates the space-value-integrated representation into action (Bast, 2011; Bast et al., 2009).

Limitations

We tested the differential functions of the hippocampal subregions in the long axis, dHP and iHP, by inactivating each subregion during goal-directed navigation. The subregional inactivation allowed us to compare the differences in navigational patterns directly between the drug conditions within subjects. However, our study includes only behavioral results and further mechanistic explanations as to the processes underlying the behavioral deficits require physiological investigations at the cellular level. Neurophysiological recordings during VR task performance could answer, for example, the questions such as whether the value-associated map in the iHP is built upon the map inherited from the dHP or it is independently developed in the iHP. Also, although our observations and behavioral data strongly suggest that rats rely on allocentric visual scenes in the VR environment instead of a single or limited set of landmarks, it is still difficult to prove experimentally whether rats used the cognitive map of the virtual arena to find the high-value zone or they had an alternative strategy to find the goal.

Materials and methods

Subjects

Eight male Long–Evans rats (8 weeks old) were housed individually under a 12 hr light/dark cycle in a temperature- and humidity-controlled environment. Rats were food-restricted to maintain ~80% of their free-feeding weight, but water was provided ad libitum. The experimental protocol (SNU-200504-3-1) complied with the guidelines of the Institutional Animal Care and Use Committee of Seoul National University. Based on our prior studies (Park et al., 2017; Yoo and Lee, 2017; Lee et al., 2014), the sample size of our study was set to the least number to achieve the necessary statistical power in the current within-subject study design for ethical commitments and practical considerations (i.e., relatively long training periods).

2D VR system

We established our own VR environment consisting of a circular arena surrounded by multiple landmarks using a game engine (Unreal Engine [UE] 4.14.3; Epic Games, Inc, USA; Figure 1A and B). The VR environment was presented via five adjacent LCD monitors covering 270° of the visual field. Rats were body-restrained at the top of a spherical treadmill, and a silicone-coated Styrofoam ball with 400 mm diameter was placed on multiple ball bearings. Rats could move their heads freely; body jackets were used to anchor their positions, limiting their body movements to a 120° range. As rats rolled the treadmill, their movement was recorded by three rotary encoders (DBS60E-BGFJD1024; Sick, Inc, Germany) attached to the treadmill surface. The signal from the encoders was then sent to the computer and synchronized with the movement in the virtual environment via an Arduino interface board (Arduino Leonardo; Arduino, Italy) and MATLAB R2021a (MathWorks, USA). A licking port was placed in front of the rats and moved in association with their body movement. It was maintained in a retracted position but was extended toward the snout by a linear motor (L16-R; Actuonix Motion Devices, Canada); an infrared sensor (FD-S32; Panasonic Industry, Japan) detected rats’ tongues to record licking behavior. When rats arrived at either reward zone, the solenoid valve (VA212-3N; Aonetech, Republic of Korea), controlled by the UE via the Arduino interface (Arduino UNO; Arduino, Italy), dispensed honey water as a reward. The amount of honey water dispensed for high-value and low-value zones was maintained at a ratio (in drops) of 12:2, with 12 μl per drop.

Behavioral paradigm

After several days of handling, rats were moved to the VR apparatus and trained to roll the treadmill to navigate the virtual environment (‘Shaping’; Figure 4B). In this session, rats had to reach a flickering checkerboard-shaped sphere randomly spawned on a circular arena (1.6 m in diameter) to obtain a honeywater reward. After rats had completed more than 60 trials on two consecutive days, they were assumed to have adapted to navigating freely. They were moved to pre-surgical training (‘Pre-training’) – a 2D VR version of the place-preference task. For the pre-training session, rats were required to find hidden reward zones using the surrounding scene, including various landmarks, such as houses, mountains, and arches, on the same circular arena from the shaping session. The start position was located at a fixed point in the center of the arena, and reward zones were located at the east and west sides of the circular platform; reward zones were positioned at a slight distance from the arena wall to prevent rats from employing a thigmotaxis strategy. Therefore, the shortest path length between the start position and the reward zone was 0.62 m. A trial started with a heading in one of six start directions, pseudorandomly chosen, and ended when the rat arrived at either reward zone. Pre-surgical training criteria were defined by the number of trials (60 trials in 40 min), high-value zone visit percentage (>75%), and average excess travel distance (<0.6 m). If a rat successfully achieved training criteria 2 days in a row, it received cannula implantation surgery.

After the surgery, rats were allowed 1 week of recovery (‘Recovery’) and then were moved to post-surgical training (‘Post-training’). During post-training, rats were tested on the same place-preference task until they achieved the same criteria as pre-surgical training, except that the trial number was reduced to 40 and the average excess travel distance was reduced to less than 1 m. This point marked the beginning of the drug injection stage; four rats received their initial injection in the dHP, and the other four rats were injected first in the iHP to counterbalance the injection order (‘Place-Preference Task’).

Object-guided navigation task

After completing drug injections, we trained five of the eight rats from the main task for an object-guided navigation task to investigate whether drug infusion caused any motor- or motivation-related impairments (‘Probe’; Figure 4B). Note the smaller sample size in the object-guided navigation task. This was because the task was later added to the study design. In this task, the rat simply had to find and navigate toward a flickering object; because there was no need for the rat to use the surrounding scene to locate the reward, this probe test was hippocampus-independent. For the probe test, the rat started from the south of the arena; concurrently, a flickering checkerboard-shaped sphere appeared on either the left or right side of the screen. When the rat arrived at the reward zone (i.e., a 0.4-m-radius circle surrounding the object), the visual stimulus stopped, a honeywater reward was given, and the trial ended. No landmarks surrounded the circular arena to distinguish the environment from that in the main task. The criterion for training included the completion of 40 trials with less than 0.1 m of mean excess travel distance, calculated as the shortest path length between the start location and the reward zone; a time limit of 60 s was also imposed. The drug infusion schedule from the place-preference task was then repeated, at which point rats were sacrificed for histological procedures.

Surgery

After rats reached pre-surgical training criteria, they were implanted with four commercial cannulae (P1 Technologies, USA), bilaterally targeting the dHP and the iHP, enabling within-subject comparisons in performance between dHP inactivation (dMUS) and iHP inactivation (iMUS) conditions (Figure 4A). Animals were first anesthetized with an intraperitoneal injection of sodium pentobarbital (Nembutal, 65 mg/kg), then their heads were fixed in a stereotaxic frame (Kopf Instruments, USA). Isoflurane (0.5–2% mixed with 100% oxygen) was used to maintain anesthesia throughout the surgery. The cannula tips targeted approximately the upper blades of the dentate gyrus of both regions (AP –3.8 mm, ML ±2.6 mm, DV –2.7 mm for the dHP; AP –6.0 mm, ML ±5.6 mm, DV –3.2 mm with a 10° tilt for the iHP) to inactivate each subregion effectively. The cannula, consisting of a 26-gauge guide cannula coupled with a 33-gauge dummy cannula, was fixed to the target location by several skull screws and bone cement. After the surgery, ibuprofen syrup was orally administered for pain relief, and the animal was kept in an intensive care unit overnight.

Drug infusion

For drug injection, the rat was first anesthetized with isoflurane. Then, 0.3–0.5 μl of either PBS or muscimol (MUS; 1 mg/ml, dissolved in saline) was infused into each hemisphere via a 33-gauge injection cannula at an injection speed of 0.167 μl/min, based on our previous study (Lee et al., 2014; Kim et al., 2012). The injection cannula and dummy cannula extended 1 mm below the tip of the guide cannula. The injection cannula was left in place for 1 min after completing the drug infusion to ensure stable diffusion of the drug. Then, it was slowly removed from the guide cannula and replaced by the dummy cannula. The rat was kept in a clean cage to recover from anesthesia completely and monitored for side effects for 20 min, then was moved to the VR apparatus for behavioral testing. If the rat showed any side effect, particularly sluggishness or aggression, we reduced the drug injection amount in the rat by 0.1 μl until we found the dosage with which there was no visible side effect. As a result, five of the rats received 0.4 μl, two received 0.3 μl, and one received 0.5 μl.

Histology

After completing the probe test, animals were sacrificed by inhalation of an overdose of CO₂. Rats were then transcardially perfused, first with PBS, administered with a syringe, and then with a 4% v/v formaldehyde solution, delivered using a commercial pump (Masterflex Easy-Load II Pump; Cole-Parmer, USA). The brain was extracted and placed in a 4% v/v formaldehyde–30% sucrose solution at 4℃ until it sank to the bottom of the container. After gelatin embedding, the brain was sectioned at 40 μm using a microtome (HM430; Thermo Fisher Scientific, USA), and sections were mounted on subbed slide glasses for Nissl staining.

Statistical analysis

Data were statistically analyzed using custom programs written in MATLAB R2021a (MathWorks), Prism 9 (GraphPad, USA), and SPSS (IBM, USA). Statistical significance was determined using the Wilcoxon signed-rank test and one-way repeated-measures analysis of variance (RM ANOVA) followed by a Bonferroni post hoc test. Although most of our statistics were based on the nonparametric tests for the relatively small sample size (n = 8), we used the parametric RM ANOVA for comparing three groups (i.e., PBS, dMUS, and iMUS) because it is the most commonly known and widely used statistical test in such comparison. However, we also performed statistical test with the alternatives for reference, and the statistical significances were not changed with any of the results. For directional analysis, Kuiper’s test and Watson–Williams test were used. However, the latter test was considered inapplicable for the mean angle when the average mean vector length between two samples was less than 0.45 (Berens, 2009). The significance level was set at α = 0.05, and all error bars indicate the standard error of means (SEMs).

Acknowledgements

This work was supported by the National Research Foundation of Korea (2019R1A2C2088799, 2021R1A4A2001803, 2022M3E5E8017723) and the Global Ph.D. Fellowship program (2019H1A2A1073456). We thank Heesoo Oh for his assistance in behavioral training.

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Contributor Information

Inah Lee, Email: inahlee@snu.ac.kr.

Laura L Colgin, University of Texas at Austin, United States.

Funding Information

This paper was supported by the following grants:

National Research Foundation of Korea 2019R1A2C2088799 to Inah Lee.
National Research Foundation of Korea 2021R1A4A2001803 to Inah Lee.
National Research Foundation of Korea 2022M3E5E8017723 to Inah Lee.
National Research Foundation of Korea 2019H1A2A1073456 to Hyeri Hwang.

Additional information

Competing interests

No competing interests declared.

Author contributions

Conceptualization, Data curation, Software, Formal analysis, Visualization, Writing – original draft, Writing – review and editing.

Resources, Software, Methodology.

Conceptualization, Resources, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing – original draft, Project administration, Writing – review and editing.

Ethics

This study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the Seoul National University. All of the animals were handled according to approved institutional animal care and use committee (IACUC) protocol (SNU-200504-3-1) of the Seoul National University.

Additional files

MDAR checklist

elife-97114-mdarchecklist1.docx^{(98.9KB, docx)}

Data availability

The behavioral data and codes used in this study can be accessed freely through https://doi.org/10.5281/zenodo.12593588.

The following dataset was generated:

Hwang H, Jin SW, Lee I. 2024. hhwang28/Hwang-et-al.-eLife-2024: Hwang et al., eLife 2024_v2. Zenodo.

References

Acharya L, Aghajan ZM, Vuong C, Moore JJ, Mehta MR. Causal influence of visual cues on hippocampal directional selectivity. Cell. 2016;164:197–207. doi: 10.1016/j.cell.2015.12.015. [DOI] [PubMed] [Google Scholar]
Ainge JA, Tamosiunaite M, Wörgötter F, Dudchenko PA. Hippocampal place cells encode intended destination, and not a discriminative stimulus, in a conditional T-maze task. Hippocampus. 2011;22:534–543. doi: 10.1002/hipo.20919. [DOI] [PubMed] [Google Scholar]
Aronov D, Tank DW. Engagement of neural circuits underlying 2D spatial navigation in a rodent virtual reality system. Neuron. 2014;84:442–456. doi: 10.1016/j.neuron.2014.08.042. [DOI] [PMC free article] [PubMed] [Google Scholar]
Bast T, Wilson IA, Witter MP, Morris RGM. From rapid place learning to behavioral performance: A key role for the intermediate hippocampus. PLOS Biology. 2009;7:e1000089. doi: 10.1371/journal.pbio.1000089. [DOI] [PMC free article] [PubMed] [Google Scholar]
Bast T. The hippocampal learning-behavior translation and the functional significance of hippocampal dysfunction in schizophrenia. Current Opinion in Neurobiology. 2011;21:492–501. doi: 10.1016/j.conb.2011.01.003. [DOI] [PubMed] [Google Scholar]
Berens P. CircStat: A MATLAB toolbox for circular statistics. Journal of Statistical Software. 2009;31:1–21. doi: 10.18637/jss.v031.i10. [DOI] [Google Scholar]
Beyeler A, Namburi P, Glober GF, Simonnet C, Calhoon GG, Conyers GF, Luck R, Wildes CP, Tye KM. Divergent routing of positive and negative information from the amygdala during memory retrieval. Neuron. 2016;90:348–361. doi: 10.1016/j.neuron.2016.03.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
Biane JS, Ladow MA, Stefanini F, Boddu SP, Fan A, Hassan S, Dundar N, Apodaca-Montano DL, Zhou LZ, Fayner V, Woods NI, Kheirbek MA. Neural dynamics underlying associative learning in the dorsal and ventral hippocampus. Nature Neuroscience. 2023;26:798–809. doi: 10.1038/s41593-023-01296-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
Bienkowski MS, Bowman I, Song MY, Gou L, Ard T, Cotter K, Zhu M, Benavidez NL, Yamashita S, Abu-Jaber J, Azam S, Lo D, Foster NN, Hintiryan H, Dong HW. Integration of gene expression and brain-wide connectivity reveals the multiscale organization of mouse hippocampal networks. Nature Neuroscience. 2018;21:1628–1643. doi: 10.1038/s41593-018-0241-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
Chen L, Lu YP, Chen HY, Huang SN, Guo YR, Zhang JY, Li QX, Luo CY, Lin SW, Chen ZN, Hu LH, Wang WX, Li HY, Cai P, Yu CX. Ventral tegmental area GABAergic neurons induce anxiety-like behaviors and promote palatable food intake. Neuropharmacology. 2020;173:108114. doi: 10.1016/j.neuropharm.2020.108114. [DOI] [PubMed] [Google Scholar]
de Hoz L, Knox J, Morris RGM. Longitudinal axis of the hippocampus: both septal and temporal poles of the hippocampus support water maze spatial learning depending on the training protocol. Hippocampus. 2003;13:587–603. doi: 10.1002/hipo.10079. [DOI] [PubMed] [Google Scholar]
Dolorfo CL, Amaral DG. Entorhinal cortex of the rat: organization of intrinsic connections. The Journal of Comparative Neurology. 1998;398:49–82. doi: 10.1002/(sici)1096-9861(19980817)398:1<49::aid-cne4>3.0.co;2-9. [DOI] [PubMed] [Google Scholar]
Dong HW, Swanson LW, Chen L, Fanselow MS, Toga AW. Genomic-anatomic evidence for distinct functional domains in hippocampal field CA1. PNAS. 2009;106:11794–11799. doi: 10.1073/pnas.0812608106. [DOI] [PMC free article] [PubMed] [Google Scholar]
Dupret D, O’Neill J, Pleydell-Bouverie B, Csicsvari J. The reorganization and reactivation of hippocampal maps predict spatial memory performance. Nature Neuroscience. 2010;13:995–1002. doi: 10.1038/nn.2599. [DOI] [PMC free article] [PubMed] [Google Scholar]
Duvelle É, Grieves RM, Hok V, Poucet B, Arleo A, Jeffery KJ, Save E. Insensitivity of place cells to the value of spatial goals in a two-choice flexible navigation task. The Journal of Neuroscience. 2019;39:2522–2541. doi: 10.1523/JNEUROSCI.1578-18.2018. [DOI] [PMC free article] [PubMed] [Google Scholar]
Felix-Ortiz AC, Tye KM. Amygdala inputs to the ventral hippocampus bidirectionally modulate social behavior. The Journal of Neuroscience. 2014;34:586–595. doi: 10.1523/JNEUROSCI.4257-13.2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
Fontanini A, Grossman SE, Figueroa JA, Katz DB. Distinct subtypes of basolateral amygdala taste neurons reflect palatability and reward. The Journal of Neuroscience. 2009;29:2486–2495. doi: 10.1523/JNEUROSCI.3898-08.2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
Gasbarri A, Packard MG, Campana E, Pacitti C. Anterograde and retrograde tracing of projections from the ventral tegmental area to the hippocampal formation in the rat. Brain Research Bulletin. 1994;33:445–452. doi: 10.1016/0361-9230(94)90288-7. [DOI] [PubMed] [Google Scholar]
Gauthier JL, Tank DW. A dedicated population for reward coding in the hippocampus. Neuron. 2018;99:179–193. doi: 10.1016/j.neuron.2018.06.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
Georges-François P, Rolls ET, Robertson RG. Spatial view cells in the primate hippocampus: allocentric view not head direction or eye position or place. Cerebral Cortex. 1999;9:197–212. doi: 10.1093/cercor/9.3.197. [DOI] [PubMed] [Google Scholar]
Gothard KM, Skaggs WE, Moore KM, McNaughton BL. Binding of hippocampal CA1 neural activity to multiple reference frames in a landmark-based navigation task. The Journal of Neuroscience. 1996;16:823–835. doi: 10.1523/JNEUROSCI.16-02-00823.1996. [DOI] [PMC free article] [PubMed] [Google Scholar]
Hollup SA, Molden S, Donnett JG, Moser MB, Moser EI. Accumulation of hippocampal place fields at the goal location in an annular watermaze task. The Journal of Neuroscience. 2001;21:1635–1644. doi: 10.1523/JNEUROSCI.21-05-01635.2001. [DOI] [PMC free article] [PubMed] [Google Scholar]
Hoover WB, Vertes RP. Anatomical analysis of afferent projections to the medial prefrontal cortex in the rat. Brain Structure & Function. 2007;212:149–179. doi: 10.1007/s00429-007-0150-4. [DOI] [PubMed] [Google Scholar]
Jarzebowski P, Hay YA, Grewe BF, Paulsen O. Different encoding of reward location in dorsal and intermediate hippocampus. Current Biology. 2022;32:834–841. doi: 10.1016/j.cub.2021.12.024. [DOI] [PubMed] [Google Scholar]
Jin SW, Lee I. Differential encoding of place value between the dorsal and intermediate hippocampus. Current Biology. 2021;31:3053–3072. doi: 10.1016/j.cub.2021.04.073. [DOI] [PubMed] [Google Scholar]
Jung MW, Wiener SI, McNaughton BL. Comparison of spatial firing characteristics of units in dorsal and ventral hippocampus of the rat. The Journal of Neuroscience. 1994;14:7347–7356. doi: 10.1523/JNEUROSCI.14-12-07347.1994. [DOI] [PMC free article] [PubMed] [Google Scholar]
Kennedy PJ, Shapiro ML. Motivational states activate distinct hippocampal representations to guide goal-directed behaviors. PNAS. 2009;106:10805–10810. doi: 10.1073/pnas.0903259106. [DOI] [PMC free article] [PubMed] [Google Scholar]
Kim S, Lee J, Lee I. The hippocampus is required for visually cued contextual response selection, but not for visual discrimination of contexts. Frontiers in Behavioral Neuroscience. 2012;6:66. doi: 10.3389/fnbeh.2012.00066. [DOI] [PMC free article] [PubMed] [Google Scholar]
Kim J, Pignatelli M, Xu S, Itohara S, Tonegawa S. Antagonistic negative and positive neurons of the basolateral amygdala. Nature Neuroscience. 2016;19:1636–1646. doi: 10.1038/nn.4414. [DOI] [PMC free article] [PubMed] [Google Scholar]
Kjelstrup KB, Solstad T, Brun VH, Hafting T, Leutgeb S, Witter MP, Moser EI, Moser MB. Finite scale of spatial representation in the hippocampus. Science. 2008;321:140–143. doi: 10.1126/science.1157086. [DOI] [PubMed] [Google Scholar]
Krettek JE, Price JL. Projections from the amygdaloid complex and adjacent olfactory structures to the entorhinal cortex and to the subiculum in the rat and cat. The Journal of Comparative Neurology. 1977;172:723–752. doi: 10.1002/cne.901720409. [DOI] [PubMed] [Google Scholar]
Lee I, Griffin AL, Zilli EA, Eichenbaum H, Hasselmo ME. Gradual translocation of spatial correlates of neuronal firing in the hippocampus toward prospective reward locations. Neuron. 2006;51:639–650. doi: 10.1016/j.neuron.2006.06.033. [DOI] [PubMed] [Google Scholar]
Lee KJ, Park SB, Lee I. Elemental or contextual? It depends: individual difference in the hippocampal dependence of associative learning for a simple sensory stimulus. Frontiers in Behavioral Neuroscience. 2014;8:217. doi: 10.3389/fnbeh.2014.00217. [DOI] [PMC free article] [PubMed] [Google Scholar]
Liu X, Carter AG. Ventral hippocampal inputs preferentially drive corticocortical neurons in the infralimbic prefrontal cortex. The Journal of Neuroscience. 2018;38:7351–7363. doi: 10.1523/JNEUROSCI.0378-18.2018. [DOI] [PMC free article] [PubMed] [Google Scholar]
Markus EJ, Qin YL, Leonard B, Skaggs WE, McNaughton BL, Barnes CA. Interactions between location and task affect the spatial and directional firing of hippocampal neurons. The Journal of Neuroscience. 1995;15:7079–7094. doi: 10.1523/JNEUROSCI.15-11-07079.1995. [DOI] [PMC free article] [PubMed] [Google Scholar]
Maurer AP, VanRhoads SR, Sutherland GR, Lipa P, McNaughton BL. Self‐motion and the origin of differential spatial scaling along the septo‐temporal axis of the hippocampus. Hippocampus. 2005;15:841–852. doi: 10.1002/hipo.20114. [DOI] [PubMed] [Google Scholar]
Morris RGM, Hagan JJ, Rawlins JNP. Allocentric spatial learning by hippocampectomised rats: a further test of the “spatial mapping” and “working memory” theories of hippocampal function. The Quarterly Journal of Experimental Psychology. 1986;38:365–395. [PubMed] [Google Scholar]
Moser MB, Moser EI, Forrest E, Andersen P, Morris RG. Spatial learning with a minislab in the dorsal hippocampus. PNAS. 1995;92:9697–9701. doi: 10.1073/pnas.92.21.9697. [DOI] [PMC free article] [PubMed] [Google Scholar]
O’Keefe J, Nadel L. The Hippocampus as a Cognitive Map. Oxford: Clarendon; 1978. [Google Scholar]
Packard MG, Hirsh R, White NM. Differential effects of fornix and caudate nucleus lesions on two radial maze tasks: evidence for multiple memory systems. The Journal of Neuroscience. 1989;9:1465–1472. doi: 10.1523/JNEUROSCI.09-05-01465.1989. [DOI] [PMC free article] [PubMed] [Google Scholar]
Park EH, Ahn JR, Lee I. Interactions between stimulus and response types are more strongly represented in the entorhinal cortex than in its upstream regions in rats. eLife. 2017;6:e32657. doi: 10.7554/eLife.32657. [DOI] [PMC free article] [PubMed] [Google Scholar]
Pikkarainen M, Rönkkö S, Savander V, Insausti R, Pitkänen A. Projections from the lateral, basal, and accessory basal nuclei of the amygdala to the hippocampal formation in rat. The Journal of Comparative Neurology. 1999;403:229–260. [PubMed] [Google Scholar]
Ravassard P, Kees A, Willers B, Ho D, Aharoni DA, Cushman J, Aghajan ZM, Mehta MR. Multisensory control of hippocampal spatiotemporal selectivity. Science. 2013;340:1342–1346. doi: 10.1126/science.1232655. [DOI] [PMC free article] [PubMed] [Google Scholar]
Rolls ET, O’Mara SM. View‐responsive neurons in the primate hippocampal complex. Hippocampus. 1995;5:409–424. doi: 10.1002/hipo.450050504. [DOI] [PubMed] [Google Scholar]
Royer S, Sirota A, Patel J, Buzsáki G. Distinct representations and theta dynamics in dorsal and ventral hippocampus. The Journal of Neuroscience. 2010;30:1777–1787. doi: 10.1523/JNEUROSCI.4681-09.2010. [DOI] [PMC free article] [PubMed] [Google Scholar]
Speakman A, O’Keefe J. Hippocampal complex spike cells do not change their place fields if the goal is moved within a cue controlled environment. The European Journal of Neuroscience. 1990;2:544–555. doi: 10.1111/j.1460-9568.1990.tb00445.x. [DOI] [PubMed] [Google Scholar]
Swanson LW, Wyss JM, Cowan WM. An autoradiographic study of the organization of intrahippocampal association pathways in the rat. The Journal of Comparative Neurology. 1978;181:681–715. doi: 10.1002/cne.901810402. [DOI] [PubMed] [Google Scholar]
Tabuchi E, Mulder AB, Wiener SI. Reward value invariant place responses and reward site associated activity in hippocampal neurons of behaving rats. Hippocampus. 2003;13:117–132. doi: 10.1002/hipo.10056. [DOI] [PubMed] [Google Scholar]
Tao S, Wang Y, Peng J, Zhao Y, He X, Yu X, Liu Q, Jin S, Xu F. Whole-brain mapping the direct inputs of dorsal and ventral CA1 projection neurons. Frontiers in Neural Circuits. 2021;15:643230. doi: 10.3389/fncir.2021.643230. [DOI] [PMC free article] [PubMed] [Google Scholar]
Tye KM, Janak PH. Amygdala neurons differentially encode motivation and reinforcement. The Journal of Neuroscience. 2007;27:3937–3945. doi: 10.1523/JNEUROSCI.5281-06.2007. [DOI] [PMC free article] [PubMed] [Google Scholar]
Valenti O, Mikus N, Klausberger T. The cognitive nuances of surprising events: exposure to unexpected stimuli elicits firing variations in neurons of the dorsal CA1 hippocampus. Brain Structure & Function. 2018;223:3183–3211. doi: 10.1007/s00429-018-1681-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
Van Groen T, Wyss JM. Connections of the retrosplenial granular b cortex in the rat. The Journal of Comparative Neurology. 2003;463:249–263. doi: 10.1002/cne.10757. [DOI] [PubMed] [Google Scholar]
Yoo SW, Lee I. Functional double dissociation within the entorhinal cortex for visual scene-dependent choice behavior. eLife. 2017;6:e21543. doi: 10.7554/eLife.21543. [DOI] [PMC free article] [PubMed] [Google Scholar]

eLife. doi: 10.7554/eLife.97114.3.sa0

eLife assessment

Laura L Colgin ¹

The authors report solid evidence for a valuable set of findings in rats performing a new virtual place-preference task. Temporary pharmacological inhibition targeting the dorsal or intermediate hippocampus disrupted navigation to a goal location in the task, and functional inhibition of the intermediate hippocampus was more detrimental than functional inhibition of the dorsal hippocampus. The work provides novel insights into functional differentiation along the dorsal-ventral axis of the hippocampus.

eLife. doi: 10.7554/eLife.97114.3.sa1

Reviewer #1 (Public review):

Anonymous

Summary:

The manuscript examines the contribution of dorsal and intermediate hippocampus to goal-directed navigation in a wide virtual environment where visual cues are provided by the scenery on the periphery of a wide arena. Among a choice of 2 reward zones located near the arena periphery, rats learn to navigate from the center of the arena to the reward zone associated with the highest reward. Navigation performance is largely assessed from the rats' body orientation when they leave the arena center and when they reach the periphery, as well as the angular mismatch between reward zone and the site rats reach the periphery. Muscimol inactivation of dorsal and intermediate hippocampus alters rat navigation to the reward zone, but the effect was more pronounced for the inactivation of intermediate hippocampus, with some rat trajectories ending in the zone associated with the lowest reward. Based on these results, the authors suggest that the intermediate hippocampus is critical especially for navigating to the highest reward zone.

Strengths:

- The authors developed an effective approach to study goal-directed navigation in a virtual environment where visual cues are provided by the peripheral scenery.

- In general, text is clearly written and the figures are well designed and relatively straightforward to interpret, even without reading the legends.

- An intriguing result, which would deserve to be better investigated and/or discussed, was that rats tended to rotate always in the counterclockwise direction. Could this be because of a hardware bias making it easier to turn left, some aspect of the peripheral landscape, or a natural preference of rats to turn left that is observable (or reported) in real environment?

- Another interesting observation, which would also deserved to be addressed in the discussion, is the fact that dHP/iHP inactivations produced to some extent consistent shifts in departing and peripheral crossing directions. This is visible from the distributions in Figures 6 and 7, which still show a peak under muscimol inactivation, but this peak is shifted to earlier angles than the correct ones. Such change is not straightforward to interpret, unlike the shortening of the mean vector length.

Maybe rats under muscimol could navigate simply using association of reward zone with some visual cues in the peripheral scene, in brain areas other than the hippocampus, and therefore stopped their rotation as soon as they saw the cues, a bit before the correct angle. While with their hippocampus intact, rats could estimate precisely the spatial relationship between the reward zone and visual cues.

Weaknesses:

- I am not sure that the differential role of dHP and iHP for navigation to high/low reward locations is supported by the data. The current results could be compatible with iHP inactivation producing a stronger impairment on spatial orientation than dHP inactivation, generating more erratic trajectories that crossed by chance the second reward zone.

To make the point that iHP inactivation affects disambiguation of high and low reward locations, the authors should show that the fraction of trajectories aiming at the low reward zone is higher than expected by chance. Somehow we would expect to see a significant peak pointing toward the low reward zone in the distribution of Figures 6-7.

Review of revised manuscript

The experimental paradigm and analyses are interesting/novel and generate some intriguing phenomena such as the animals' preference for counterclockwise rotation and the stereotypical trajectory shifts induced by muscimol inactivation. Understanding better the underlying mechanisms of these phenomena and the navigational strategies involved in this apparatus will be important in the future for correctly interpreting inactivation experiments.

The idea of a differential effect of dMUS and iMUS was toned down in the abstract and other parts of the manuscript, such that the claims now better match the data.

eLife. doi: 10.7554/eLife.97114.3.sa2

Reviewer #2 (Public review):

Anonymous

Summary:

The aim of this paper was to elucidate the role of the dorsal HP and intermediate HP (dHP and iHP) in value-based spatial navigation through behavioral and pharmacological experiments using a newly developed VR apparatus. The authors inactivated dHP and iHP by muscimol injection and analyzed the differences in behavior. The results showed that dHP was important for spatial navigation, while iHP was critical for both value judgments and spatial navigation. The present study developed a new sophisticated behavioral experimental apparatus and proposed a behavioral paradigm that is useful for studying value-dependent spatial navigation. In addition, the present study provides important results that support previous findings of differential function along the dorsoventral axis of the hippocampus.

eLife. doi: 10.7554/eLife.97114.3.sa3

Reviewer #3 (Public review):

Anonymous

Summary:

The authors established a new virtual reality place preference task. On the task, rats, which were body-restrained on top of a moveable Styrofoam ball and could move through a circular virtual environment by moving the Styrofoam ball, learnt to navigate reliably to a high-reward location over a low-reward location, using allocentric visual cues arranged around the virtual environment.

The authors also showed that functional inhibition by bilateral microinfusion of the GABA-A receptor agonist muscimol, which targeted the dorsal or intermediate hippocampus, disrupted task performance. The impact of functional inhibition targeting the intermediate hippocampus was more pronounced than that of functional inhibition targeting the dorsal hippocampus.

Moreover, the authors demonstrated that the same manipulations did not significantly disrupt rats' performance on a virtual reality task that required them to navigate to a spherical landmark to obtain reward, although there were numerical impairments in the main performance measure and the absence of statistically significant impairments may partly reflect a small sample size (see under Weaknesses, point 3.).

Overall, the study established a new virtual-reality place preference task for rats and established that performance on this task requires the dorsal to intermediate hippocampus. It also established that task performance is more sensitive to the same muscimol infusion when the infusion was applied to the intermediate hippocampus, compared to the dorsal hippocampus. The authors suggest that these differential effects of muscimol infusions reflect that dorsal hippocampus is responsible for 'precise' spatial navigation and intermediate hippocampus for place-value associations, but this idea remains to be tested by further studies. In their first revision to the paper, the authors toned down this claim, but I still think it would be good to consider more explicitly potential alternative explanations for the differential effects of dorsal and intermediate muscimol infusions (see under Weaknesses, point 2.).

Strengths:

(1) The authors established a new place preference task for body-restrained rats in a virtual environment and, using temporary pharmacological inhibition by intra-cerebral microinfusion of the GABA-A receptor agonist muscimol, showed that task performance requires dorsal to intermediate hippocampus.

(2) These findings extend our knowledge about place learning tasks that require dorsal to intermediate hippocampus and add to previous evidence that the intermediate hippocampus may be more important than other parts of the hippocampus, including the dorsal hippocampus, for goal-directed navigation based on allocentric place memory.

(3) The hippocampus-dependent task may be useful for future recording studies examining how hippocampal neurons may support behavioral performance based on place information.

Weaknesses:

(1) The new findings do not strongly support the authors' suggestion that dorsal hippocampus is responsible for precise spatial navigation and intermediate hippocampus for place-value associations (e.g., final sentence of the first paragraph of the Discussion). The authors base this claim on differential effects of the dorsal and intermediate hippocampal muscimol infusions on different performance measures on the virtual reality place preference task. More specifically, dorsal hippocampal muscimol infusion significantly increased perimeter crossings and perimeter crossing deviations, whereas other measures of task performance are not significantly changed, including departure direction and visits to the high-value location. However, these statistical outcomes offer only limited evidence that dorsal hippocampal infusion specifically affected the perimeter crossing, without affecting the other measures. Numerically the pattern of infusion effects is quite similar across these various measures: intermediate hippocampal infusions markedly impaired these performance measures compared to vehicle infusions, and the values of these measures after dorsal hippocampal muscimol infusion were between the values in the intermediate hippocampal muscimol and the vehicle condition (Figs 5-7). In my opinion, these findings could reflect that dorsal and intermediate hippocampus play distinct roles, as suggested by the authors, but the findings are also consistent with the suggestion that intermediate hippocampal muscimol had a quantitatively stronger, but qualitatively similar effect to dorsal hippocampal muscimol. However, I am largely content with the authors acknowledging within the paper that their suggestion would need to be confirmed by additional studies.

Moreover, I do not find it clearly described in the paper which distinct aspects of navigation the departure direction and perimeter crossing deviation measures capture. The authors suggest that departure direction and perimeter crossing deviation are indices of the navigational efficiency and precision of navigation, respectively (e.g., from p. 7, line 195). However, both departure direction and perimeter crossing deviation measure how accurate/precise, in other words 'close to the target', the rat's navigation is. Efficiency of navigation may rather be reflected by the path length taken (a measure that was not reported). It appears to me that a key difference between the two measures is that departure direction measures the rat's direction towards the goal at the beginning of the rat's navigational path, whereas perimeter crossing deviation measures this further toward the end of the navigational path. This would suggest that departure direction may depend more on directional orienting mechanisms early on in the rat's journey, whereas perimeter crossing deviation may also depend on fine-grained place recognition as the rat approaches the goal. Given the fine-grained place representations in the dorsal hippocampus, the latter may, therefore, depend more on the dorsal hippocampus than the former. I think this would fit with the authors' suggestion 'that the dHP represents a fine-scaled spatial map of an environment' (p. 18, first line). If the authors agree with my interpretation of the different measures, they may consider clarifying this in the Results and Discussion sections.

(2) The claim that the different effects of intermediate and dorsal hippocampal muscimol infusions reflect different functions of intermediate and dorsal hippocampus rests on the assumption that both manipulations inhibit similar volumes of hippocampal tissue to a similar extent, but at different levels along the dorso-ventral axis of the hippocampus. However, this is not a foregone conclusion (e.g., drug spread may differ depending on the infusion site or drug effects may differ due to differential distribution or efficiency of GABA-A receptors), and the authors do not provide direct evidence for this assumption. Therefore, an alternative account of the weaker effects of dorsal compared to intermediate hippocampal muscimol infusions on place-preference performance is that the dorsal infusions affect less hippocampal volume or less markedly inhibit neurons within the affected volume than the intermediate infusions (e.g., due to different drug spread following dorsal and intermediate infusions or due to different distribution or effectiveness of GABA-A receptors in dorsal and intermediate hippocampus). I would recommend that the authors explicitly state this limitation in the limitations section of the Discussion. In their response to my original comments, the authors argue that it is unlikely that muscimol exerts stronger effects in intermediate compared to dorsal hippocampus, based on the finding that in vitro paired pulse inhibition is reduced in ventral compared to dorsal hippocampal slices (Papatheodoropoulos et al., 2002). However, this claim is not strongly supported by the in vitro paired-pulse inhibition findings. First, these findings relate to differences between dorsal and ventral hippocampus, whereas differences between dorsal and intermediate hippocampus were not investigated. Second, reduced paired pulse inhibition may not necessarily reflect reduced GABA-A receptor expression/efficiency (which would be likely to reduce muscimol effects), but may also reflect reduced GABAergic input, which would not be expected to reduce muscimol effects.

(3) It is good that the authors included a comparison/control experiment using a spherical beacon-guided navigation task, to examine the specific psychological mechanisms disrupted by the hippocampal manipulations. However, the sample size for the comparison experiment (n = 5 rats) was lower than for the main study n = 8 rats, and the data shown in Fig. 8 suggest that the comparison task may be affected by the hippocampal manipulations similarly to the place-preference task, albeit less markedly. This effect may well have been significant if the same sample size had been used as in the main experiment. Therefore, I would recommend that the authors acknowledge this limitation in the Discussion (perhaps, in the Limitation section).

eLife. 2024 Jul 16;13:RP97114. doi: 10.7554/eLife.97114.3.sa4

Author response

Hyeri Hwang ¹, Seung-Woo Jin ², Inah Lee ³

The following is the authors’ response to the original reviews.

Public Reviews:

Reviewer #1 (Public Review):

Summary:

The manuscript examines the contribution of the dorsal and intermediate hippocampus to goal-directed navigation in a wide virtual environment where visual cues are provided by the scenery on the periphery of a wide arena. Among a choice of 2 reward zones located near the arena periphery, rats learn to navigate from the center of the arena to the reward zone associated with the highest reward. Navigation performance is largely assessed from the rats' body orientation when they leave the arena center and when they reach the periphery, as well as the angular mismatch between the reward zone and the site rats reach the periphery. Muscimol inactivation of the dorsal and intermediate hippocampus alters rat navigation to the reward zone, but the effect was more pronounced for the inactivation of the intermediate hippocampus, with some rat trajectories ending in the zone associated with the lowest reward. Based on these results, the authors suggest that the intermediate hippocampus is critical, especially for navigating to the highest reward zone.

Strengths:

-The authors developed an effective approach to study goal-directed navigation in a virtual environment where visual cues are provided by the peripheral scenery.

- In general, the text is clearly written and the figures are well-designed and relatively straightforward to interpret, even without reading the legends.

- An intriguing result, which would deserve to be better investigated and/or discussed, was that rats tended to rotate always in the counterclockwise direction. Could this be because of a hardware bias making it easier to turn left, some aspect of the peripheral landscape, or a natural preference of rats to turn left that is observable (or reported) in a real environment?

Thank you for the insightful question. As the reviewer mentioned, the counterclockwise rotation behavior was intriguing and unexpected. To answer the reviewer’s question properly, we examined whether such stereotypical turning behavior appeared before the rats acquired the task rule and reward zones in the pre-surgical training phase of the task. Data from the last day of shaping and the first day of the pre-surgical main task day showed no significant difference in the number of trials in which the first body-turn was either clockwise or counterclockwise, suggesting that the rats did not have a bias toward a specific side (p=0.46 for Shaping; p=0.76 for the Main task, Wilcoxon signed-rank test). These results excluded the possibility that there was something in the apparatus's hardware that made the rats turn only to the left. Also, since we used the same peripheral landscape for the shaping and main task, we could assume that the peripheral landscape did not cause movement bias.

Although it remains inconclusive, we have noticed that some prior studies alluded to a phenomenon similar to this issue, framed as the topic of lateralization or spatial preference by comparing left and right biases. For example, Wishaw et al. (1992) suggested that there was natural lateralization in rats (“Most of the rats displayed either a strong right limb bias or a strong left limb bias.”) but no dominance to a specific side. Andrade et al. (2001) also claimed that “83% of Wistar rats spontaneously showed a clear preference for left or right arms in the T-maze.” However, to the best of our knowledge, there has been no direct evidence that rats have a dominant natural preference only to one side.

Therefore, while the left-turning behavior remains an intriguing topic for further investigation, we find it difficult to pinpoint the reason behind the behavior in the current study. However, we would like to emphasize that this behavior did not interrupt testing our hypothesis. Nonetheless, we agree with the reviewer’s point that the counterclockwise rotation needs to be discussed more, so we revised the manuscript as follows:

“To rule out the potential effect of hardware bias or any particular aspect of peripheral landscape to make rats turn only to one side, we measured the direction of the first body-turn in each trial on the last day of shaping and the first day of the main task (i.e., before rats learned the reward zones). There was no significant difference between the clockwise and counterclockwise turns (p=0.46 for shaping, p=0.76 for main task; Wilcoxon signed-rank test), indicating that the stereotypical pattern of counterclockwise body-turn appeared only after the rats learned the reward locations.” (p.6)

- Another interesting observation, which would also deserve to be addressed in the discussion, is the fact that dHP/iHP inactivations produced to some extent consistent shifts in departing and peripheral crossing directions. This is visible from the distributions in Figures 6 and 7, which still show a peak under muscimol inactivation, but this peak is shifted to earlier angles than the correct ones. Such change is not straightforward to interpret, unlike the shortening of the mean vector length.

Maybe rats under muscimol could navigate simply by using the association of reward zone with some visual cues in the peripheral scene, in brain areas other than the hippocampus, and therefore stopped their rotation as soon as they saw the cues, a bit before the correct angle. While with their hippocampus is intact, rats could estimate precisely the spatial relationship between the reward zone and visual cues.

We agree with the possibility suggested by the reviewer. However, although not described in the original manuscript, we performed several different control experiments in a few rats using various visual stimulus manipulations to test how their behaviors change as a result. One of the experiments was the landmark omission test, where one of the landmarks was omitted. The landmark to be made disappear was pseudorandomly manipulated on a trial-by-trial basis. We observed that the omission of one landmark, regardless of its identity, did not cause a specific behavioral change in finding the reward zones, suggesting that the rats were not relying on a single visual landmark when finding the reward zone.

Therefore, it is unlikely that rats used the spatial relationship between the reward zone and a specific visual cue to solve the task in our study. However, the result was based on an insufficient sample size (n = 3), not permitting any meaningful statistical testing. Thus, we have now updated this information in the manuscript as an anecdotal result as follows:

“Additionally, to investigate whether the rats used a certain landmark as a beacon to find the reward zones, we conducted the landmark omission test as a part of control experiments. Here, one of the landmarks was omitted, and the landmark to be made disappear was pseudorandomly manipulated on a trial-by-trial basis. The omission of one landmark, regardless of its identity, did not cause a specific behavioral change in finding the reward zones, suggesting that the rats were not relying on a single visual landmark when finding the reward zones. The result can be reported anecdotally only because of an insufficient sample size (n = 3), not permitting any meaningful statistical testing.” (p.9)

Weaknesses:

-I am not sure that the differential role of dHP and iHP for navigation to high/low reward locations is supported by the data. The current results could be compatible with iHP inactivation producing a stronger impairment on spatial orientation than dHP inactivation, generating more erratic trajectories that crossed by chance the second reward zone.

To make the point that iHP inactivation affects the disambiguation of high and low reward locations, the authors should show that the fraction of trajectories aiming at the low reward zone is higher than expected by chance. Somehow we would expect to see a significant peak pointing toward the low reward zone in the distribution of Figures 6-7.

We thank the reviewer for the valuable comments. We agree that it is difficult to rigorously distinguish the loss of value representation from spatial disorientation in our experiment. Since the trial ended once the rat touched either reward zone, it was difficult to specify whether they intended to arrive at the location or just moved randomly and arrived there by chance. Moreover, it is possible that the drug infusion did not completely inactivate the iHP but only partially did so.

To investigate this issue further, we checked whether the distribution of the departure direction (DD) differed between the trials in which rats initially headed north (NW, N, NE) and south (SE, S, SW) at the start. In the manuscript, we demonstrated that DD aligned with the high-value zone, indicating that the rat remembered the scenes associated with the high-value zone (p.8). Based on the rats’ characteristic counterclockwise rotation, the reward zone rats would face first upon starting while heading north would be the high-value zone. On the other hand, the rat would face the low-value reward zone when starting while heading south. In this case, normal rats would inhibit leaving the start zone and rotate further until they face the high-value zone before finally departing the start location. If the iHP inactivation caused a more severe impairment in spatial orientation but not in value representation, it is likely that the iHP-inactivated rats in both north- and south-starting trials would behave similarly with the dHP-inactivated rats, but producing a larger deviation from the high-value zone. However, if the iHP inactivation affected the disambiguation of high and low reward locations, north and south-starting trials would show different DD distributions.

The circular plots shown below are the DD distributions of dMUS and iMUS. We could see that when they started facing north, iHP-inactivated rats still aligned themselves towards the high-value zone and thus remained spatially oriented, similar to the dHP inactivation session. However, in the south-starting trials, the DD distribution was completely different from the north-starting trials; the rats failed in body alignment towards the high-value zone. Instead, they departed the start point while heading south in most trials. This pattern was not seen in dMUS sessions, even in their south-starting trials, illustrating the distinct deficit caused by iHP inactivation. Additionally, most of the rats with iHP inactivation visited the low-value zone more in south-headed starting trials than in the north-headed trials, except for one rat.

Furthermore, we would like to clarify that we do not limit the effect of iHP inactivation to the impairment in distinguishing the high and low reward zones. It is possible that iHP inactivation resulted in the loss of a global value-representing map, leading to the impairment in distinguishing both reward zones from other non-rewarded areas in the environment. Figures 6 and 7 implicated this possibility by showing that the peaks are not restricted only to the reward zones. Unfortunately, we cannot rigorously address this in the current study because of the limitations of our experimental design mentioned above.

Nonetheless, we agree with the reviewer that this limitation needs to be addressed, so we now added how the current study needs further investigation to clarify what causes the behavioral change after the iHP inactivation in the Limitations section (p.21).

Reviewer #2 (Public Review):

Summary:

The aim of this paper was to elucidate the role of the dorsal HP and intermediate HP (dHP and iHP) in value-based spatial navigation through behavioral and pharmacological experiments using a newly developed VR apparatus. The authors inactivated dHP and iHP by muscimol injection and analyzed the differences in behavior. The results showed that dHP was important for spatial navigation, while iHP was critical for both value judgments and spatial navigation. The present study developed a new sophisticated behavioral experimental apparatus and proposed a behavioral paradigm that is useful for studying value-dependent spatial navigation. In addition, the present study provides important results that support previous findings of differential function along the dorsoventral axis of the hippocampus.

Strengths:

The authors developed a VR-based value-based spatial navigation task that allowed separate evaluation of "high-value target selection" and "spatial navigation to the target." They were also able to quantify behavioral parameters, allowing detailed analysis of the rats' behavioral patterns before and after learning or pharmacological inactivation.

Weaknesses:

Although differences in function along the dorsoventral axis of the hippocampus is an important topic that has received considerable attention, differences in value coding have been shown in previous studies, including the work of the authors; the present paper is an important study that supports previous studies, but the novelty of the findings is not that high, as the results are from pharmacological and behavioral experiments only.

We appreciate the reviewer's insightful comments. In response, we would like to emphasize that a very limited number of studies investigated the function of the intermediate hippocampus, especially in spatial memory tasks. We tested the differential functions of the dorsal and intermediate hippocampus using a within-animal design and used reversible inactivation manipulation (i.e., muscimol injection) to prevent potential compensation by other brain regions when using irreversible manipulation techniques (i.e., lesion). Also, very few studies have analyzed the navigation trajectories of animals as closely as in the current study. We emphasize the novelty of our study by comparing it with prior studies, as shown below in Table 1.

Author response table 1. Comparison of our study with those from prior studies.

	Target regions	Within animal comparison	Manipulation	Value difference	Detailed,analysis,on,trajectory
Jarzebowski et al., 2022	Dorsal, intermediate	No	No	No	No
Bast et al., 2009	Dorsal, intermediate, ventral	No	Lesion	No	No
Moser et al., 1995	Dorsal, ventral	No	Lesion	No	No
De Hoz et al., 2003	Dorsal, ventral	No	Lesion	No	No
Ferbinteanu and McDonald, 2001	Dorsal, ventral	No	Lesion	No	Yes (heading angle)
De Saint Blanquat et al., 2013	Dorsal, intermediate- ventral	No	Reversible inactivation (Muscimol)	No	No
Tabuchi et al., 2003	Dorsal, ventral	Yes	No	Yes	No
Jin and Lee, 2021 (previous study)	Dorsal, intermediate	Yes	No	Yes	No
Hwang et al., 2024 (current study)	Dorsal, intermediate	Yes	Reversible inactivation (Muscimol)	Yes	Yes

Open in a new tab

Moreover, to the best of our knowledge, the current manuscript is the first to investigate the hippocampal subregions along the long axis in a VR environment using a hippocampal-dependent spatial memory task. Nonetheless, we agree that the current study has a limitation as a behavior-only experiment. We now have added a comment on how other techniques, such as electrophysiology, would develop our findings in the Limitation section (p.21).

Reviewer #3 (Public Review):

Summary:

The authors established a new virtual reality place preference task. On the task, rats, which were body-restrained on top of a moveable Styrofoam ball and could move through a circular virtual environment by moving the Styrofoam ball, learned to navigate reliably to a high-reward location over a low-reward location, using allocentric visual cues arranged around the virtual environment.

The authors also showed that functional inhibition by bilateral microinfusion of the GABA-A receptor agonist muscimol, which targeted the dorsal or intermediate hippocampus, disrupted task performance. The impact of functional inhibition targeting the intermediate hippocampus was more pronounced than that of functional inhibition targeting the dorsal hippocampus.

Moreover, the authors demonstrated that the same manipulations did not significantly disrupt rats' performance on a virtual reality task that required them to navigate to a spherical landmark to obtain reward, although there were numerical impairments in the main performance measure and the absence of statistically significant impairments may partly reflect a small sample size (see comments below).

Overall, the study established a new virtual-reality place preference task for rats and established that performance on this task requires the dorsal to intermediate hippocampus. They also established that task performance is more sensitive to the same muscimol infusion (presumably - doses and volumes used were not clearly defined in the manuscript, see comments below) when the infusion was applied to the intermediate hippocampus, compared to the dorsal hippocampus, although this does not offer strong support for the authors claim that dorsal hippocampus is responsible for accurate spatial navigation and intermediate hippocampus for place-value associations (see comments below).

Strengths:

(1) The authors established a new place preference task for body-restrained rats in a virtual environment and, using temporary pharmacological inhibition by intra-cerebral microinfusion of the GABA-A receptor agonist muscimol, showed that task performance requires dorsal to intermediate hippocampus.

(2) These findings extend our knowledge about place learning tasks that require dorsal to intermediate hippocampus and add to previous evidence that, for some place memory tasks, the intermediate hippocampus may be more important than other parts of the hippocampus, including the dorsal hippocampus, for goal-directed navigation based on allocentric place memory.

(3) The hippocampus-dependent task may be useful for future recording studies examining how hippocampal neurons support behavioral performance based on place information.

Weaknesses:

(1) The new findings do not strongly support the authors' suggestion that the dorsal hippocampus is responsible for accurate spatial navigation and the intermediate hippocampus for place-value associations.

The authors base this claim on the differential effects of the dorsal and intermediate hippocampal muscimol infusions on different performance measures. More specifically, dorsal hippocampal muscimol infusion significantly increased perimeter crossings and perimeter crossing deviations, whereas dorsal infusion did not significantly change other measures of task performance, including departure direction and visits to the high-value location. However, these statistical outcomes offer only limited evidence that dorsal hippocampal infusion specifically affected the perimeter crossing, without affecting the other measures. Numerically the pattern of infusion effects is quite similar across these various measures: intermediate hippocampal infusions markedly impaired these performance measures compared to vehicle infusions, and the values of these measures after dorsal hippocampal muscimol infusion were between the values in the intermediate hippocampal muscimol and the vehicle condition (Figures 5-7). Moreover, I am not so sure that the perimeter crossing measures really reflect distinct aspects of navigational performance compared to departure direction and hit rate, and, even if they did, which aspects this would be. For example, in line 316, the authors suggest that 'departure direction and PCD [perimeter crossing deviation] [are] indices of the effectiveness and accuracy of navigation, respectively'. However, what do the authors mean by 'effectiveness' and 'accuracy'? Accuracy typically refers to whether or not the navigation is 'correct', i.e. how much it deviates from the goal location, which would be indexed by all performance measures.

So, overall, I would recommend toning down the claim that the findings suggest that the dorsal hippocampus is responsible for accurate spatial navigation and the intermediate hippocampus for place-value associations.

The reviewer mentioned that the statistical outcomes offer limited evidence as the dHP inactivation results were always positioned between the results of the iHP inactivation and controls. However, we would like to emphasize that, projecting to each other, the two subregions are not completely segregated anatomically. It is highly likely this is also true functionally and there should be some overlap in their roles. Considering such relationships between the dHP and iHP, it could be natural to see an intermediate effect after inactivating the dHP, and that is why we focused on the “magnitude” of behavioral changes after inactivation instead of complete dissociation between the two subregions in our manuscript. Unfortunately, because of the nature of the drug infusion study, further dissociation would be difficult, requiring further investigation with different experimental techniques, such as physiological examinations of the neural firing patterns between the two regions. We mentioned this caveat of the current study in the Limitations as follows:

“However, our study includes only behavioral results and further mechanistic explanations as to the processes underlying the behavioral deficits require physiological investigations at the cellular level. Neurophysiological recordings during VR task performance could answer, for example, the questions such as whether the value-associated map in the iHP is built upon the map inherited from the dHP or it is independently developed in the iHP.” (p.21)

Regarding the reviewer’s comment on the meaning of measuring the perimeter crossing directions, we would like to draw the reviewer’s attention to the individual trajectories during the iMUS sessions described in Figure 5. Particularly when they were not confident with the location of the higher reward, rats changed their heading directions during the navigation, which resulted in a less efficient route to the goal location. Rats showing this type of behavior tended to hit the perimeter of the arena first before correcting their routes toward the goal zone. In contrast, rats showing effective navigation hardly bumped into the wall or perimeter before hitting the goal zone. Thus, their PCDs matched DDs almost always. When considered together with DD, our PCD measure could tell whether rats not hitting the goal zone directly after departure were impaired in either maintaining the correct heading direction to the goal zone at the start location or orienting themselves to the target zone accurately from the start. Our results suggest that the latter is the case. We included the relevant explanation in the Discussion section as follows:

“Particularly, rats changed their heading directions during the navigation when they were not confident with the location of the higher reward, resulting in a less efficient route to the goal location. Rats showing this type of behavior tended to hit the perimeter of the arena first before correcting their routes. Therefore, when considered together with DD, our PCD measure could tell that the rats not hitting the goal zone directly after departure were impaired in orienting themselves to the target zone accurately from the start, not in maintaining the correct heading direction to the goal zone at the start location.” (p.19)

Nonetheless, we agree with the reviewer that the term ‘accuracy’ might be confusing with performance accuracy, so we replaced the term with ‘precision’ throughout the manuscript, referring to the precise targeting of the reward zones.

(2) The claim that the different effects of intermediate and dorsal hippocampal muscimol infusions reflect different functions of intermediate and dorsal hippocampus rests on the assumption that both manipulations inhibit similar volumes of hippocampal tissue to a similar extent, but at different levels along the dorso-ventral axis of the hippocampus. However, this is not a foregone conclusion (e.g., drug spread may differ depending on the infusion site or drug effects may differ due to differential expression of GABA-A receptors in the dorsal and intermediate hippocampus), and the authors do not provide direct evidence for this assumption. Therefore, a possible alternative account of the weaker effects of dorsal compared to intermediate hippocampal muscimol infusions on place-preference performance is that the dorsal infusions affect less hippocampal volume or less markedly inhibit neurons within the affected volume than the intermediate infusions. I would recommend that the authors briefly consider this issue in the discussion. Moreover, from the Methods, it is not clear which infusion volume and muscimol concentration were used for the different infusions (see below, 4.a.), and this must be clarified.

We appreciate these insightful comments from the reviewer and agree that we do not provide direct evidence for the point raised by the reviewer. To the best of our knowledge, most of the behavioral studies on the long axis of the hippocampus did not particularly address the differential expression of GABA-A receptors along the axis. We could not find any literature that specifically introduced and compared the levels of expression of GABA-A receptors or the diffusion range of muscimol in the intermediate hippocampus to the other subregions. However, we found that Sotiriou et al. (2005) made such comparisons with respect to the expression of different GABA-A receptors. They concluded that the dorsal and ventral hippocampi have different levels of the GABA-A receptor subtypes. The a1/b2/g2 subtype was dominant in the dorsal hippocampus, while the a2/b1/g2 subtype was prevalent in the ventral hippocampus. Sotiriou and colleagues also mentioned the lower affinity of GABA-A receptor binding in the ventral hippocampus, and this result is consistent with the Papatheodoropoulos et al. (2002) study that showed a weaker synaptic inhibition in the ventral hippocampus compared to the dorsal hippocampus. Papatheodoropoulos et al. speculated differences in GABA receptors as one of the potential causes underlying the differential synaptic inhibition between the dorsal and ventral hippocampal regions. Based on these findings, the same volume of muscimol is more likely to cause a more severe effect on the ventral hippocampus than the dorsal hippocampus. Therefore, we do not believe that the less significant changes after the dorsal hippocampal inactivation were induced by the expression level of GABA-A receptors. Additionally, we have demonstrated in our previous study that muscimol injections in the dorsal hippocampus impair performance to the chance level in scene-based behavioral tasks (Lee et al., 2014; Kim et al., 2012).

Nonetheless, we mentioned the possibility of differential muscimol expressions between the two target regions. Following the suggestion of the reviewer, we now included this information in the Discussion as follows:

“Although there is still a possibility that the levels of expression of GABA-A receptors might be different along the longitudinal axis of the hippocampus, …” (p.20)

Regarding the drug infusion volume and concentration, we included these details in the Methods. Please see our detailed response to 4.a. below.

(3) It is good that the authors included a comparison/control study using a spherical beacon-guided navigation task, to examine the specific psychological mechanisms disrupted by the hippocampal manipulations. However, as outlined below (4.b.), the sample size for the comparison study was lower than for the main study, and the data in Figure 8 suggest that the comparison task may be affected by the hippocampal manipulations similarly to the place-preference task, albeit less markedly. This would raise the question as to which mechanisms that are common to the two tasks may be affected by hippocampal functional inhibition, which should be considered in the discussion.

The sample size for the object-guided navigation task was smaller because we initially did not plan the experiment, but later in the study decided to conduct the control test. Therefore, the object-guided navigation task was added to the study design after finishing the first three rats, resulting in a smaller sample size than the place preference task. We included this detail in the manuscript, as follows:

“Note the smaller sample size in the object-guided navigation task. This was because the task was later added to the study design.” (p.24)

Regarding the mechanism behind the two different tasks, we did not perform the same heading direction analysis here as in the place preference task because the two tasks have different characteristics such as task complexity. The object-guided navigation task is somewhat similar to the visually guided (or cued) version of the water maze task, which is widely known as hippocampal-independent (Morris et al., 1986; Packard et al., 1989; also see our descriptions on p.15). Therefore, we would argue that the two tasks (i.e., place preference task and object-guided navigation task) used in the current manuscript do not share neural mechanisms in common. Additionally, we confirmed that several behavioral measurements related to motor capacity, such as travel distance and latency, along with the direct hit proportion provided in Figure 8, did not show any statistically significant changes across drug conditions.

4. Several important methodological details require clarification:

a. Drug infusions (from line 673):

- '0.3 to 0.5 μl of either phosphate-buffered saline (PBS) or muscimol (MUS) was infused into each hemisphere'; the authors need to clarify when which infusion volume was used and why different infusion volumes were used.

We thank the reviewer for carefully reading our manuscript. We were cautious about side effects, such as suppressed locomotion or overly aggressive behavior, since the iHP injection site was close to the ventricle. We were keenly aware that the intermediate to ventral hippocampal regions are sensitive to the drug dosage from our previous experiments. Thus, we observed the rat’s behavior for 20 minutes after drug injection in a clean cage. We started from 0.5 μl, based on our previous study, but if the injected rat showed any sign of side effects in the cage, we stopped the experiment for the day and tried with a lower dosage (i.e., 0.4 μl first, then 0.3 μl, etc.) until we found the right dosage under which the rat did not show any side effect. This procedure is necessary because cannula tip positions are slightly different from rat to rat. When undergoing this procedure, five out of eight rats received 0.4 μl, two received 0.3 μl, and one received 0.5 μl. Still, there was no significant difference in performance, including the high-value visit percentage, departing and perimeter crossing directions, across all dosages. This information is now added in the Methods section as follows:

“If the rat showed any side effect, particularly sluggishness or aggression, we reduced the drug injection amount in the rat by 0.1 ml until we found the dosage with which there was no visible side effect. As a result, five of the rats received 0.4 ml, two received 0.3 ml, and one received 0.5 ml.” (p.25)

- I could not find the concentration of the muscimol solution that was used. The authors must clarify this and also should include a justification of the doses used, e.g. based on previous studies.

Thank you for the suggestion. We used the drug concentration of 1mg/ml, which was adapted from our previous muscimol study (Lee et al., 2014; Kim et al., 2012). The manuscript is now updated, as follows:

“…or muscimol (MUS; 1mg/ml, dissolved in saline) was infused into each hemisphere via a 33-gauge injection cannula at an injection speed of 0.167 ml/min, based on our previous study (Lee et al., 2014; Kim et al., 2012).” (p.25)

- Please also clarify if the injectors and dummies were flush with the guides or by which distance they protruded from the guides.

The injection and dummy cannula both protruded from the guide cannula by 1 mm, and this information is now added to the Methods section, as follows:

“The injection cannula and dummy cannula extended 1 mm below the tip of the guide cannula.” (p.25)

b. Sample sizes: The authors should include sample size justifications, e.g. based on considerations of statistical power, previous studies, practical considerations, or a combination of these factors. Importantly, the smaller sample size in the control study using the spherical beacon-guided navigation task (n = 5 rats) limits comparability with the main study using the place-preference task (n = 8). Numerically, the findings on the control task (Figure 8) look quite similar to the findings on the place-preference task, with intermediate hippocampal muscimol infusions causing the most pronounced impairment and dorsal hippocampal muscimol infusions causing a weaker impairment. These effects may have reached statistical significance if the same sample size had been used in the place-preference study.

We set the current sample size for several reasons. First, based on our previous studies, we assumed that eight, or more than six, would be enough to achieve statistical power in a “within-animal design” study. Also, considering the ethical commitments, we tried to keep the number of animals used in the study to the least. Last, our paradigm required very long training periods (3 months on average per animal), so we could not increase the sample size for practical reasons. Regarding the reasons for the smaller sample size for the object-guided navigation task, please see the previous response to 3 above. The manuscript is now revised as follows:

“Based on our prior studies (Park et al., 2017; Yoo and Lee, 2017; Lee et al., 2014), the sample size of our study was set to the least number to achieve the necessary statistical power in the current within-subject study design for ethical commitments and practical considerations (i.e., relatively long training periods).” (p.22)

c. Statistical analyses: Why were the data of the intermediate and dorsal hippocampal PBS infusion conditions averaged for some of the analyses (Figure 5; Figure 6B and C; Figure 7B and C; Figure 8B) but not for others (Figure 6A and Figure 7A)?

The reviewer is correct that we only illustrated the separate dPBS and iPBS data for Figures 6A and 7A. Since the directional analysis is the main focus of the current manuscript, we tried to provide better visualization and more detailed examples of how the drug infusion changed the behavioral patterns between the PBS and MUS conditions in each region. Except for the visualization of DD and PCD, we averaged the PBS sessions to increase statistical power, as described in p.9. We added a detailed description of the reasons for illustrating dPBS and iPBS data separately in the manuscript, as follows:

“Note that dPBS and iPBS sessions were separately illustrated here for better visualization of changes in the behavioral pattern for each subregion.” (p.12)

Reviewing Editor (Recommendations For The Authors):

The strength of evidence rating in the assessment is currently noted as "incomplete." This can be improved following revisions if you amend your conclusions in the paper, including in the title and abstract, such that the paper's major conclusions more closely match what is shown in the Results.

Following the suggestions of the reviewing editor, we have mentioned the caveats of our study in the Limitations section of our revised manuscript (p.21). In addition, the manuscript has been revised so that the conclusions in the paper match more closely to the experimental results as can been seen in some of the relevant sentences in the abstract and main text as follows:

“Inactivation of both dHP and iHP with muscimol altered efficiency and precision of wayfinding behavior, but iHP inactivation induced more severe damage, including impaired place preference. Our findings suggest that the iHP is more critical for value-dependent navigation toward higher-value goal locations.” (Abstract; p.2)

“Whereas inactivation of the dHP mainly affected the precision of wayfinding, iHP inactivation impaired value-dependent navigation more severely by affecting place preference.” (p.5)

“The iHP causes more damage to value-dependent spatial navigation than the dHP, which is important for navigational precision” (p.12)

However, we haven’t changed the title of the manuscript as it carries what we’d like to deliver in this study accurately.

Reviewer #1 (Recommendations For The Authors):

- What were the dimensions of the environment? What distance did rats typically run to reach the reward zone? A scale bar would be helpful in Figure 1.

We used the same circular arena from the shaping session, which was 1.6 meters in diameter (p.23), and the shortest path between the start location and either reward zone was 0.62 meters. We revised the manuscript for clarification as follows:

“For the pre-training session, rats were required to find hidden reward zones…, on the same circular arena from the shaping session.” (p.23)

“Therefore, the shortest path length between the start position and the reward zone was 0.62 meters.” (p.23)

We also added a scale bar in Figure 1C for a better understanding.

- Line 169: "The scene rotation plot covers the period from the start of the trial to when the rat leaves the starting point at the center and the departure circle (Figure 2B)."

The sentence is unclear. Maybe it should be "... from the start of the trial to when the rat leaves the departure circle”.

The sentence has been revised following the reviewer's suggestion. (p.7)

- Line 147: "First, they learned to rotate the spherical treadmill counterclockwise to move around in the virtual environment (presumably to perform energy-efficient navigation)."

It is not clear from this sentence if rats naturally preferred the counterclockwise direction or if the counterclockwise direction was a task requirement.

We now clarified in our revised manuscript that it was not a task requirement to turn counterclockwise, as follows:

“First, although it was not required in the task, they learned to rotate the spherical treadmill counterclockwise…” (p.6)

- Line 149: "Second, once a trial started, but before leaving the starting point at the center, the animal rotated the treadmill to turn the virtual environment immediately to align its starting direction with the visual scene associated with the high-value reward zone."

The sentence is unclear. Maybe "Second, once a trial started, the animal rotated the treadmill immediately to align its starting direction with the visual scene associated with the high-value reward zone.”

We have updated the description following the suggestion. (p.6)

Reviewer #2 (Recommendations For The Authors):

- There are some misleading descriptions of the conclusion of the results in this paper. In this study, the functions of (a) selection of high-value target and (b) spatial navigation to the target were assessed in the behavioral experiments. The results of the pharmacological experiments showed that dHP inactivation impaired (b) and iHP inactivation impaired both (a) and (b) (Figures 5 B & D). However, the last sentence of the abstract states that dHP is important for the functions of (a) and iHP for (b). There are several other similar statements in the main text. Since the separation of (a) and (b) is an important and original aspect of this study, the description should clearly show the conclusion that dHP is important for (a) and iHP is important for both (a) and (b).

Related to the above, the paragraph title in the Discussion "The iHP may contain a value-associated cognitive map with reasonable spatial resolution for goal-directed navigation (536-537)" is also somewhat misleading: "with reasonable resolution for goal-directed behavior" seems to reflect the results of an object-guided navigation task (Figure 8). However, the term "goal-directed behavior" is also used for value-dependent spatial navigation (i.e., the main task), which causes confusion. I would like to suggest clarifying the wording on this point.

First, we need to correct the reviewer’s statement regarding our descriptions of the results. As the reviewer mentioned, our results indicated that the dHP inactivation impaired (b) but not (a), while the iHP inactivation impaired both (a) and (b). Regarding the iHP inactivation result, we focused on the impairment of (a) since our aim was to investigate spatial-value association in the hippocampus. Also, it was more likely that (a) affected (b), but not the other way, because (a) remained intact when (b) was impaired after dHP inactivation. We emphasized this difference between dHP and iHP inactivation, which was (a). Therefore, we mentioned in the last sentence of the abstract that the dHP is important for (b), which is the precision of spatial navigation to the target location, and the iHP is critical for (a).

Moreover, we would like to clarify that we were not referring to the object-guided navigation task in Figure 8 in the phrase ‘with a reasonable spatial resolution for goal-directed navigation.’ Please note that the object-guided navigation task did not require fine spatial resolution to find the reward. The phrase instead referred to the dHP inactivation result (Figure 5 and 6), where the rats could find the high-value zone even with dHP inactivation, although the navigational precision decreased. Nonetheless, we agree with the reviewer for the confusion that the title might cause, so now have updated the title as follows:

“The iHP may contain a value-associated cognitive map with reasonable spatial resolution for value-based navigation” (p.19)

- As an earlier study focusing on the physiology of iHP, Maurer et al, Hippocampus 15:841 (2005) is also a pioneering and important study, and I suggest citing it.

Thank you for the suggestion. We included the Maurer et al. (2005) study in the Introduction section as follows:

“…Specifically, there is physiological evidence that the size of a place field becomes larger as recordings of place cells move from the dHP to the vHP (Jung et al., 1994; Maurer et al., 2005; Kjelstrup et al., 2008; Royer et al., 2010).” (p.4)

- One of the strengths of this paper is that we have developed a new control system for the VR navigation task device, but I cannot get a very detailed description of this system in the Methods section. Also, no information about the system control has been uploaded to GitHub. I would suggest adding a description of the manufacturer, model number, and size of components, such as a rotary encoder and ball, and information about the software of the control system, with enough detail to allow the reader to reconstruct the system.

We have now added detailed descriptions of the VR system in the Methods section (see “2D VR system). (p.22)

Reviewer #3 (Recommendations For The Authors):

(1) Some comments on specific passages of text:

Lines 87 to 89: 'Surprisingly, beyond the recognition of anatomical divisions, little is known about the functional differentiation of subregions along the dorsoventral axis of the hippocampus. Moreover, the available literature on the subject is somewhat inconsistent.'

I would recommend to rephrase these statements. Regarding the first statement, there is substantial evidence for functional differentiation along the dorso-ventral axis of the hippocampus (e.g., see reviews by Moser and Moser, 1998, Hippocampus; Bannerman et al., 2004, Neurosci Biobehav Rev; Bast, 2007, Rev Neurosci; Bast, 2011, Curr Opin Neurobiol; Fanselow and Dong, 2010, Neuron; Strange et al., 2014, Nature Rev Neurosci). Regarding the second statement, the authors may consider being more specific, as the inconsistencies demonstrated seem to relate mainly to the hippocampal representation of value information, instead of functional differentiation along the dorso-ventral hippocampal axis in general.

We agree with the reviewer that the abovementioned statements need further clarification. The manuscript is now revised as follows:

“Surprisingly, beyond the recognition of anatomical divisions, the available literature on the functional differentiation of subregions along the dorsoventral axis of the hippocampus, particularly in the context of value representation, is somewhat inconsistent.” (p.4)

Lines 92 to 93: 'Thus, it has been thought that the dHP is more specialized for precise spatial representation than the iHP and vHP.'

I think 'fine-grained' may be the more appropriate term here. Also, check throughout the manuscript when referring to the differences of spatial representations along the hippocampal dorso-ventral axis.

Thank you for the insightful suggestion. We changed the term to ‘fine-grained’ throughout the manuscript, as follows:

“Thus, it has been thought that the dHP is more specialized for fine-grained spatial representation than the iHP and vHP.” (p.4)

“Consequently, the fine-grained spatial map present in the dHP…” (p.20)

Line 217: well-'trained' rats?

We initially used the term ‘well-learned’ to focus on the effect of learning, not training. Please note that the rats were already adapted to moving freely in the VR environment during the Shaping sessions, but the immediate counterclockwise body alignment only appeared after they acquired the reward locations for the main task. Nonetheless, we agree that the term might cause confusion, so we revised the manuscript as the reviewer suggested, as follows:

“This implies that well-trained rats aligned their bodies more efficiently…” (p.8)

Lines 309 to 311: 'Taken together, these results indicate that iHP inactivation severely damages normal goal-directed navigational patterns in our place preference task.'

Consider to mention that dHP inactivation also causes impairments, albeit weaker ones.

We thank the reviewer for the suggestion. We revised the manuscript by mentioning dHP inactivation as follows:

“Taken together, these results indicate that iHP inactivation more severely damages normal goal-directed navigational patterns than dHP inactivation in our place-preference task.” (p.11-12)

Lines 550 to 552: 'The involvement of the iHP in spatial value association has been reported in several studies. For example, Bast and colleagues reported that rapid place learning is disrupted by removing the iHP and vHP, even when the dHP remains undamaged (Bast et al., 2009).'

Bast et al. (2009) did not directly show the role of iHP in 'spatial value associations'. They suggested that the importance of iHP for behavioral performance based on rapid, one-trial, place learning may reflect neuroanatomical features of the intermediate region, especially the combination of afferents that could convey the required fine-grained visuo-spatial information with relevant afferent and efferent connections that may be important to translate hippocampal place memory into appropriate behavioral performance (this may include afferents conveying value information). More recent theoretical and empirical research suggests that projections to the (ventral) striatum may be relevant (see Tessereau et al., 2021, BNA and Bauer et al., 2021, BNA).

We appreciate the reviewer for this insightful comment. We agree with the reviewer that Bast et al. (2009) did not directly mention spatial value association; however, learning a new platform location needs an update of value information in the spatial environment. Therefore, we thought the study, though indirectly, suggested how the iHP contributes to spatial value associations. Nonetheless, to avoid confusion, we revised the manuscript, as follows:

“The involvement of the iHP in spatial value association has been reported or implicated in several studies” (p.20)

(2) Figures and legends:

Figure 2B: What do the numbers after novice and expert indicate?

The numbers indicate the rat ID, followed by the session number. We added the details to the Figure legend, as follows:

“The numbers after ‘Novice’ and ‘Expert’ indicate the rat and session number of the example.” (p.34)

Figure 2C: Please indicate units of the travel distance and latency measurements.

The units are now described in the Figure legends, as follows:

“Mean travel distance in meters and latency in seconds are shown below the VR arena trajectory.” (p.34)

Figure 3Aii: Here and in other figures - do the vector lengths have a unit (degree?)?

No, the mean vector length is an averaged value of the resultant vectors, thus having no specific unit.

Figure 5A: Please explain what the numbers on top of the individual sample trajectories indicate.

The numbers are IDs for rats, sessions, and trials of specific examples. We added the explanation to the Figure legends, as follows:

“Numbers above each trajectory indicate the identification numbers for rat, session, and trial.” (p.35)

(3) Additional comments on some methodological details:

a. Why was the non-parametric Wilcoxon signed-rank test used for the planned comparison between intermediate and dorsal hippocampal PBS infusions, whereas parametric ANOVA and post-hoc comparisons were used for other analyses? This probably doesn't make a big difference for the interpretation of the present data (as a parametric pairwise comparison would also not have revealed any significant difference between intermediate and dorsal hippocampal PBS infusions), but it would nevertheless be good to clarify the rationale for this.

We used the non-parametric statistics since our sample size was rather small (n = 8) to use the parametric statistics, although we used the parametric ANOVA for some of the results because it is the most commonly known and widely used statistical test in such comparisons. However, we also checked the statistics with the alternatives (i.e., non-parametric Wilcoxon signed-rank test to parametric paired t-test and parametric One-way RM ANOVA with Bonferroni post hoc test to non-parametric Friedman’s test with Dunn’s post hoc test), and the statistical significance did not change with any of the tests. We now added the explanation in the manuscript, as follows:

“Although most of our statistics were based on the non-parametric tests for the relatively small sample size (n = 8), we used the parametric RM ANOVA for comparing three groups (i.e., PBS, dMUS, and iMUS) because it is the most commonly known and widely used statistical test in such comparison. However, we also performed statistical tests with the alternatives for reference, and the statistical significances were not changed with any of the results.” (p.26)

b. Single housing of rats:

Why was this chosen? Based on my experience, this is not necessary for studies involving cannula implants and food restriction. Group housing is generally considered to improve the welfare of rats.

We chose single housing of rats because our training paradigm required precise restrictions on the food consumption of individual rats, which could be difficult in group housing.

c. Anesthesia:

Why was pentobarbital used, alongside isoflurane, to anesthetize rats for surgery (line 663)? The use of gaseous anesthesia alone offers very good control of anesthesia and reduces the risk of death from anesthesia compared to the use of pentobarbital.

Why was anesthesia used for the drug infusions (line 674)? If rats are well-habituated to handling by the experimenter, manual restraint is sufficient for intra-cerebral infusions. Therefore, anesthesia could be omitted, reducing the risk of adverse effects on the experimental rats.

I do not think that points b. and c. are relevant for the interpretation of the present findings, but the authors may consider these points for future studies to improve further the welfare of the experimental rats.

We appreciate the reviewer’s careful suggestions. For both the use of pentobarbital during surgery and anesthesia for the drug infusion, we chose to do so to avoid any risk of rats being awake and becoming anxious and to ensure safety during the procedures. They might not be necessary, but they were helpful for the experimenters to proceed with sufficient time to maintain precision. Nonetheless, we agree with the reviewer’s concern, which was the reason why we monitored the rats’ behavior for 20 minutes in the cage after drug infusion to minimize any potential influence on the task performance. We updated the relevant details in the Methods section, as follows:

“The rat was kept in a clean cage to recover from anesthesia completely and monitored for side effects for 20 minutes, then was moved to the VR apparatus for behavioral testing.” (p.25)

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Citations

Hwang H, Jin SW, Lee I. 2024. hhwang28/Hwang-et-al.-eLife-2024: Hwang et al., eLife 2024_v2. Zenodo. [DOI]

Supplementary Materials

MDAR checklist

elife-97114-mdarchecklist1.docx^{(98.9KB, docx)}

Data Availability Statement

The behavioral data and codes used in this study can be accessed freely through https://doi.org/10.5281/zenodo.12593588.

The following dataset was generated:

Hwang H, Jin SW, Lee I. 2024. hhwang28/Hwang-et-al.-eLife-2024: Hwang et al., eLife 2024_v2. Zenodo.

[bib1] Acharya L, Aghajan ZM, Vuong C, Moore JJ, Mehta MR. Causal influence of visual cues on hippocampal directional selectivity. Cell. 2016;164:197–207. doi: 10.1016/j.cell.2015.12.015. [DOI] [PubMed] [Google Scholar]

[bib2] Ainge JA, Tamosiunaite M, Wörgötter F, Dudchenko PA. Hippocampal place cells encode intended destination, and not a discriminative stimulus, in a conditional T-maze task. Hippocampus. 2011;22:534–543. doi: 10.1002/hipo.20919. [DOI] [PubMed] [Google Scholar]

[bib3] Aronov D, Tank DW. Engagement of neural circuits underlying 2D spatial navigation in a rodent virtual reality system. Neuron. 2014;84:442–456. doi: 10.1016/j.neuron.2014.08.042. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib4] Bast T, Wilson IA, Witter MP, Morris RGM. From rapid place learning to behavioral performance: A key role for the intermediate hippocampus. PLOS Biology. 2009;7:e1000089. doi: 10.1371/journal.pbio.1000089. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib5] Bast T. The hippocampal learning-behavior translation and the functional significance of hippocampal dysfunction in schizophrenia. Current Opinion in Neurobiology. 2011;21:492–501. doi: 10.1016/j.conb.2011.01.003. [DOI] [PubMed] [Google Scholar]

[bib6] Berens P. CircStat: A MATLAB toolbox for circular statistics. Journal of Statistical Software. 2009;31:1–21. doi: 10.18637/jss.v031.i10. [DOI] [Google Scholar]

[bib7] Beyeler A, Namburi P, Glober GF, Simonnet C, Calhoon GG, Conyers GF, Luck R, Wildes CP, Tye KM. Divergent routing of positive and negative information from the amygdala during memory retrieval. Neuron. 2016;90:348–361. doi: 10.1016/j.neuron.2016.03.004. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib8] Biane JS, Ladow MA, Stefanini F, Boddu SP, Fan A, Hassan S, Dundar N, Apodaca-Montano DL, Zhou LZ, Fayner V, Woods NI, Kheirbek MA. Neural dynamics underlying associative learning in the dorsal and ventral hippocampus. Nature Neuroscience. 2023;26:798–809. doi: 10.1038/s41593-023-01296-6. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib9] Bienkowski MS, Bowman I, Song MY, Gou L, Ard T, Cotter K, Zhu M, Benavidez NL, Yamashita S, Abu-Jaber J, Azam S, Lo D, Foster NN, Hintiryan H, Dong HW. Integration of gene expression and brain-wide connectivity reveals the multiscale organization of mouse hippocampal networks. Nature Neuroscience. 2018;21:1628–1643. doi: 10.1038/s41593-018-0241-y. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib10] Chen L, Lu YP, Chen HY, Huang SN, Guo YR, Zhang JY, Li QX, Luo CY, Lin SW, Chen ZN, Hu LH, Wang WX, Li HY, Cai P, Yu CX. Ventral tegmental area GABAergic neurons induce anxiety-like behaviors and promote palatable food intake. Neuropharmacology. 2020;173:108114. doi: 10.1016/j.neuropharm.2020.108114. [DOI] [PubMed] [Google Scholar]

[bib11] de Hoz L, Knox J, Morris RGM. Longitudinal axis of the hippocampus: both septal and temporal poles of the hippocampus support water maze spatial learning depending on the training protocol. Hippocampus. 2003;13:587–603. doi: 10.1002/hipo.10079. [DOI] [PubMed] [Google Scholar]

[bib12] Dolorfo CL, Amaral DG. Entorhinal cortex of the rat: organization of intrinsic connections. The Journal of Comparative Neurology. 1998;398:49–82. doi: 10.1002/(sici)1096-9861(19980817)398:1<49::aid-cne4>3.0.co;2-9. [DOI] [PubMed] [Google Scholar]

[bib13] Dong HW, Swanson LW, Chen L, Fanselow MS, Toga AW. Genomic-anatomic evidence for distinct functional domains in hippocampal field CA1. PNAS. 2009;106:11794–11799. doi: 10.1073/pnas.0812608106. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib14] Dupret D, O’Neill J, Pleydell-Bouverie B, Csicsvari J. The reorganization and reactivation of hippocampal maps predict spatial memory performance. Nature Neuroscience. 2010;13:995–1002. doi: 10.1038/nn.2599. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib15] Duvelle É, Grieves RM, Hok V, Poucet B, Arleo A, Jeffery KJ, Save E. Insensitivity of place cells to the value of spatial goals in a two-choice flexible navigation task. The Journal of Neuroscience. 2019;39:2522–2541. doi: 10.1523/JNEUROSCI.1578-18.2018. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib16] Felix-Ortiz AC, Tye KM. Amygdala inputs to the ventral hippocampus bidirectionally modulate social behavior. The Journal of Neuroscience. 2014;34:586–595. doi: 10.1523/JNEUROSCI.4257-13.2014. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib17] Fontanini A, Grossman SE, Figueroa JA, Katz DB. Distinct subtypes of basolateral amygdala taste neurons reflect palatability and reward. The Journal of Neuroscience. 2009;29:2486–2495. doi: 10.1523/JNEUROSCI.3898-08.2009. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib18] Gasbarri A, Packard MG, Campana E, Pacitti C. Anterograde and retrograde tracing of projections from the ventral tegmental area to the hippocampal formation in the rat. Brain Research Bulletin. 1994;33:445–452. doi: 10.1016/0361-9230(94)90288-7. [DOI] [PubMed] [Google Scholar]

[bib19] Gauthier JL, Tank DW. A dedicated population for reward coding in the hippocampus. Neuron. 2018;99:179–193. doi: 10.1016/j.neuron.2018.06.008. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib20] Georges-François P, Rolls ET, Robertson RG. Spatial view cells in the primate hippocampus: allocentric view not head direction or eye position or place. Cerebral Cortex. 1999;9:197–212. doi: 10.1093/cercor/9.3.197. [DOI] [PubMed] [Google Scholar]

[bib21] Gothard KM, Skaggs WE, Moore KM, McNaughton BL. Binding of hippocampal CA1 neural activity to multiple reference frames in a landmark-based navigation task. The Journal of Neuroscience. 1996;16:823–835. doi: 10.1523/JNEUROSCI.16-02-00823.1996. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib22] Hollup SA, Molden S, Donnett JG, Moser MB, Moser EI. Accumulation of hippocampal place fields at the goal location in an annular watermaze task. The Journal of Neuroscience. 2001;21:1635–1644. doi: 10.1523/JNEUROSCI.21-05-01635.2001. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib23] Hoover WB, Vertes RP. Anatomical analysis of afferent projections to the medial prefrontal cortex in the rat. Brain Structure & Function. 2007;212:149–179. doi: 10.1007/s00429-007-0150-4. [DOI] [PubMed] [Google Scholar]

[bib24] Jarzebowski P, Hay YA, Grewe BF, Paulsen O. Different encoding of reward location in dorsal and intermediate hippocampus. Current Biology. 2022;32:834–841. doi: 10.1016/j.cub.2021.12.024. [DOI] [PubMed] [Google Scholar]

[bib25] Jin SW, Lee I. Differential encoding of place value between the dorsal and intermediate hippocampus. Current Biology. 2021;31:3053–3072. doi: 10.1016/j.cub.2021.04.073. [DOI] [PubMed] [Google Scholar]

[bib26] Jung MW, Wiener SI, McNaughton BL. Comparison of spatial firing characteristics of units in dorsal and ventral hippocampus of the rat. The Journal of Neuroscience. 1994;14:7347–7356. doi: 10.1523/JNEUROSCI.14-12-07347.1994. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib27] Kennedy PJ, Shapiro ML. Motivational states activate distinct hippocampal representations to guide goal-directed behaviors. PNAS. 2009;106:10805–10810. doi: 10.1073/pnas.0903259106. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib28] Kim S, Lee J, Lee I. The hippocampus is required for visually cued contextual response selection, but not for visual discrimination of contexts. Frontiers in Behavioral Neuroscience. 2012;6:66. doi: 10.3389/fnbeh.2012.00066. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib29] Kim J, Pignatelli M, Xu S, Itohara S, Tonegawa S. Antagonistic negative and positive neurons of the basolateral amygdala. Nature Neuroscience. 2016;19:1636–1646. doi: 10.1038/nn.4414. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib30] Kjelstrup KB, Solstad T, Brun VH, Hafting T, Leutgeb S, Witter MP, Moser EI, Moser MB. Finite scale of spatial representation in the hippocampus. Science. 2008;321:140–143. doi: 10.1126/science.1157086. [DOI] [PubMed] [Google Scholar]

[bib31] Krettek JE, Price JL. Projections from the amygdaloid complex and adjacent olfactory structures to the entorhinal cortex and to the subiculum in the rat and cat. The Journal of Comparative Neurology. 1977;172:723–752. doi: 10.1002/cne.901720409. [DOI] [PubMed] [Google Scholar]

[bib32] Lee I, Griffin AL, Zilli EA, Eichenbaum H, Hasselmo ME. Gradual translocation of spatial correlates of neuronal firing in the hippocampus toward prospective reward locations. Neuron. 2006;51:639–650. doi: 10.1016/j.neuron.2006.06.033. [DOI] [PubMed] [Google Scholar]

[bib33] Lee KJ, Park SB, Lee I. Elemental or contextual? It depends: individual difference in the hippocampal dependence of associative learning for a simple sensory stimulus. Frontiers in Behavioral Neuroscience. 2014;8:217. doi: 10.3389/fnbeh.2014.00217. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib34] Liu X, Carter AG. Ventral hippocampal inputs preferentially drive corticocortical neurons in the infralimbic prefrontal cortex. The Journal of Neuroscience. 2018;38:7351–7363. doi: 10.1523/JNEUROSCI.0378-18.2018. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib35] Markus EJ, Qin YL, Leonard B, Skaggs WE, McNaughton BL, Barnes CA. Interactions between location and task affect the spatial and directional firing of hippocampal neurons. The Journal of Neuroscience. 1995;15:7079–7094. doi: 10.1523/JNEUROSCI.15-11-07079.1995. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib36] Maurer AP, VanRhoads SR, Sutherland GR, Lipa P, McNaughton BL. Self‐motion and the origin of differential spatial scaling along the septo‐temporal axis of the hippocampus. Hippocampus. 2005;15:841–852. doi: 10.1002/hipo.20114. [DOI] [PubMed] [Google Scholar]

[bib37] Morris RGM, Hagan JJ, Rawlins JNP. Allocentric spatial learning by hippocampectomised rats: a further test of the “spatial mapping” and “working memory” theories of hippocampal function. The Quarterly Journal of Experimental Psychology. 1986;38:365–395. [PubMed] [Google Scholar]

[bib38] Moser MB, Moser EI, Forrest E, Andersen P, Morris RG. Spatial learning with a minislab in the dorsal hippocampus. PNAS. 1995;92:9697–9701. doi: 10.1073/pnas.92.21.9697. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib39] O’Keefe J, Nadel L. The Hippocampus as a Cognitive Map. Oxford: Clarendon; 1978. [Google Scholar]

[bib40] Packard MG, Hirsh R, White NM. Differential effects of fornix and caudate nucleus lesions on two radial maze tasks: evidence for multiple memory systems. The Journal of Neuroscience. 1989;9:1465–1472. doi: 10.1523/JNEUROSCI.09-05-01465.1989. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib41] Park EH, Ahn JR, Lee I. Interactions between stimulus and response types are more strongly represented in the entorhinal cortex than in its upstream regions in rats. eLife. 2017;6:e32657. doi: 10.7554/eLife.32657. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib42] Pikkarainen M, Rönkkö S, Savander V, Insausti R, Pitkänen A. Projections from the lateral, basal, and accessory basal nuclei of the amygdala to the hippocampal formation in rat. The Journal of Comparative Neurology. 1999;403:229–260. [PubMed] [Google Scholar]

[bib43] Ravassard P, Kees A, Willers B, Ho D, Aharoni DA, Cushman J, Aghajan ZM, Mehta MR. Multisensory control of hippocampal spatiotemporal selectivity. Science. 2013;340:1342–1346. doi: 10.1126/science.1232655. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib44] Rolls ET, O’Mara SM. View‐responsive neurons in the primate hippocampal complex. Hippocampus. 1995;5:409–424. doi: 10.1002/hipo.450050504. [DOI] [PubMed] [Google Scholar]

[bib45] Royer S, Sirota A, Patel J, Buzsáki G. Distinct representations and theta dynamics in dorsal and ventral hippocampus. The Journal of Neuroscience. 2010;30:1777–1787. doi: 10.1523/JNEUROSCI.4681-09.2010. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib46] Speakman A, O’Keefe J. Hippocampal complex spike cells do not change their place fields if the goal is moved within a cue controlled environment. The European Journal of Neuroscience. 1990;2:544–555. doi: 10.1111/j.1460-9568.1990.tb00445.x. [DOI] [PubMed] [Google Scholar]

[bib47] Swanson LW, Wyss JM, Cowan WM. An autoradiographic study of the organization of intrahippocampal association pathways in the rat. The Journal of Comparative Neurology. 1978;181:681–715. doi: 10.1002/cne.901810402. [DOI] [PubMed] [Google Scholar]

[bib48] Tabuchi E, Mulder AB, Wiener SI. Reward value invariant place responses and reward site associated activity in hippocampal neurons of behaving rats. Hippocampus. 2003;13:117–132. doi: 10.1002/hipo.10056. [DOI] [PubMed] [Google Scholar]

[bib49] Tao S, Wang Y, Peng J, Zhao Y, He X, Yu X, Liu Q, Jin S, Xu F. Whole-brain mapping the direct inputs of dorsal and ventral CA1 projection neurons. Frontiers in Neural Circuits. 2021;15:643230. doi: 10.3389/fncir.2021.643230. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib50] Tye KM, Janak PH. Amygdala neurons differentially encode motivation and reinforcement. The Journal of Neuroscience. 2007;27:3937–3945. doi: 10.1523/JNEUROSCI.5281-06.2007. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib51] Valenti O, Mikus N, Klausberger T. The cognitive nuances of surprising events: exposure to unexpected stimuli elicits firing variations in neurons of the dorsal CA1 hippocampus. Brain Structure & Function. 2018;223:3183–3211. doi: 10.1007/s00429-018-1681-6. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib52] Van Groen T, Wyss JM. Connections of the retrosplenial granular b cortex in the rat. The Journal of Comparative Neurology. 2003;463:249–263. doi: 10.1002/cne.10757. [DOI] [PubMed] [Google Scholar]

[bib53] Yoo SW, Lee I. Functional double dissociation within the entorhinal cortex for visual scene-dependent choice behavior. eLife. 2017;6:e21543. doi: 10.7554/eLife.21543. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Differential functions of the dorsal and intermediate regions of the hippocampus for optimal goal-directed navigation in VR space

Hyeri Hwang

Seung-Woo Jin

Inah Lee

Roles

Abstract

Introduction

Results

Well-trained rats align themselves toward the high-value zone before departure in the place-preference task

Figure 1. Place-preference task in a 2D virtual reality (VR) environment.

Figure 2. Common body-turning behavior of rats after learning.

Departing orientation and perimeter-crossing direction provide a measure of navigational efficiency and precision, respectively

Figure 3. Learning index for efficient navigation during pre-surgical training.

Navigation is impaired by inactivation of either the dHP or iHP, but only iHP inactivation affects place-preference behavior

Figure 4. Cannula implantation locations and schedules for training and drug injection.

Figure 5. Changes in navigational pattern with each drug condition.

The iHP causes more damage to value-dependent spatial navigation than the dHP, which is important for navigational precision

Figure 6. Dorsal hippocampal (dHP) and intermediate hippocampal (iHP) inactivation differentially affect efficient goal-directed navigation.

Figure 7. Precision of goal-directed navigation is more severely impaired with intermediate hippocampal (iHP) inactivation.

Hippocampal inactivation does not impair cue-guided navigation or goal-directedness

Figure 8. Goal-directedness and navigational capacity are unaffected by drug infusion.

Discussion

Rats use allocentric visual scenes and landmarks to target the goal zone and adjust their paths accordingly during navigation in the VR environment

Inactivating the dHP impairs navigational precision but does not affect place preference based on differential reward values

The iHP may contain a value-associated cognitive map with reasonable spatial resolution for value-based navigation

Limitations

Materials and methods

Subjects

2D VR system

Behavioral paradigm

Object-guided navigation task

Surgery

Drug infusion

Histology

Statistical analysis

Acknowledgements

Funding Statement

Contributor Information

Funding Information

Additional information

Competing interests

Author contributions

Ethics

Additional files

Data availability

References

eLife assessment

Laura L Colgin

Roles

Reviewer #1 (Public review):

Anonymous

Roles

Reviewer #2 (Public review):

Anonymous

Roles

Reviewer #3 (Public review):

Anonymous

Roles

Author response

Hyeri Hwang

Seung-Woo Jin

Inah Lee

Roles

Author response image 1.

Author response image 2.

Author response image 3.

Author response table 1. Comparison of our study with those from prior studies.

Associated Data

Data Citations

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases