Abstract

Representing an environment globally, in a coarse way, and locally, in a fine-grained way, are two fundamental aspects of how our brain interprets the world that surrounds us. The neural correlates of these representations have not been explicated in humans. In this study we used fMRI to investigate these correlates and to explore a possible functional segregation in the hippocampus and parietal cortex. We hypothesized that processing a coarse, global environmental representation engages anterior parts of these regions, whereas processing fine-grained, local environmental information engages posterior parts. Participants learned a virtual environment and then had to find their way during fMRI. After scanning, we assessed strategies used and representations stored. Activation in the hippocampal head (anterior) was related to the multiple distance and global direction judgments and to the use of a coarse, global environmental representation during navigation. Activation in the hippocampal tail (posterior) was related to both local and global direction judgments and to using strategies like number of turns. A structural shape analysis showed that the use of a coarse, global environmental representation was related to larger right hippocampal head volume and smaller right hippocampal tail volume. In the inferior parietal cortex, a similar functional segregation was observed, with global routes represented anteriorly and fine-grained route information such as number of turns represented posteriorly. In conclusion, moving from the anterior to the posterior hippocampus and inferior parietal cortex reflects a shift from processing coarse global environmental representations to processing fine-grained, local environmental representations.

INTRODUCTION

Recent research indicates that space is processed along the entire axis of the hippocampus, but that the spatial representations differ in their nature. Findings in rodents demonstrate that hippocampal place fields increase in size when moving from posterior (dorsal) to anterior (ventral) regions and that different firing fields overlap more in the anterior than in the posterior hippocampus (Kjelstrup et al., 2008; Jung, Wiener, & McNaughton, 1994). These findings suggest that the anterior hippocampus forms a coarse, global representation of the environment, whereas the posterior hippocampus forms fine-grained, local representations.

Recent fMRI studies in humans support this idea. We have found that retrieving a coarse, global representation of an episode's spatial location activates the anterior hippocampus, whereas retrieving information about fine-grained spatial relations within the episode, for example, the seating arrangement of the guests in a wedding, activates the posterior hippocampus (Nadel, Hoscheidt, & Ryan, 2012). In another fMRI study, participants were shown pictures of familiar landmarks, and activation in the hippocampal head was found to increase with increasing real-world distance between landmarks (Morgan, MacEvoy, Aguirre, & Epstein, 2011). However, no human study has yet made a direct comparison between fine-grained, local and coarse, global spatial processing in the hippocampus.

The proposed distinction between anterior and posterior hippocampal representations of the environment may be a general principle of functional organization in the brain. In the visual system, the size and complexity of receptive fields increase from posterior to anterior regions, both within the occipital lobe and across other parts of the brain (Serences & Yantis, 2006). In rodents, parietal neurons have been observed to fire according to a route-centered reference frame during navigation (Nitz, 2006, 2009). It is possible that, in the human parietal cortex, the most complete route representations are processed anteriorly (Serences & Yantis, 2006), since this region is part of the dorsal visual stream (Silver & Kastner, 2009). However, this possibility remains unexplored.

A typical environmental navigation period can be divided into initial self-localization within the environment (self-localization), followed by planning how to get to the target landmark (planning), and finally moving to the target landmark (execution). We have previously shown that self-localization and planning result in a massive increase in anterior hippocampal activation, as well as posterior hippocampal activation, compared with the execution period (Xu, Evensmoen, Lehn, Pintzka, & Håberg, 2010). Additionally, self-localization and planning, compared with execution, have been shown to be characterized by retrieval of environmental representations (Xu et al., 2010; Spiers & Maguire, 2006). Thus, the hippocampus appears to be particularly important for the self-localization and planning periods of navigation, likely reflecting retrieval of environmental representations. The aim of this study was to investigate how spatial representations differ along the anterior–posterior axis of the human hippocampus. We also explored a possible segregation of route representations in the parietal cortex. To this end, we used fMRI of navigation; specifically comparing self-localization and planning with execution, in a learned virtual environment. The nature of representations acquired and used was evaluated by environmental tests and a questionnaire, respectively. Finally, hippocampal morphology was analyzed. We hypothesized that (a) activation and shape of the hippocampal head correlate with the use of a coarse, global environmental representation; (b) activation and shape of the hippocampal tail correlate with the use of fine-grained, local environmental representations; and (c) in the parietal cortex, posterior activation correlates with the use of fine-grained, local route representations and anterior activation correlates with the use of coarse, global route representations.

METHODS

Participants

Thirty men (19–28 years, mean = 22.8 years) participated in this study, all right-handed as ascertained with the Edinburg Handedness Inventory (mean score = 89.7 ± 11.7%). We included only men to reduce the number of confounding factors, because sex differences have sometimes, but not always, been observed in navigation (Barra, Laou, Poline, Lebihan, & Berthoz, 2012; Ohnishi, Matsuda, Hirakata, & Ugawa, 2006; Blanch, Brennan, Condon, Santosh, & Hadley, 2004; Grön, Wunderlich, Spitzer, Tomczak, & Riepe, 2000). The participants had no history of neurological disorders, head trauma, or current DSM-IV axis I diagnosis of psychiatric illness including substance abuse. All participants provided written informed consent before participation and received 500 Norwegian kroner as reimbursement. The study was approved by the Regional Committee for Medical Research Ethics in Midt-Norge, Norway, and adhered to the Declaration of Helsinki.

Virtual Reality Environment

The virtual reality (VR) environment was developed in collaboration with Terra Vision AS (Terra Vision, Trondheim, Norway) using the Torque game engine (Garage Games, Eugene, OR). The environment is 115.28 by 138.46 units in size, with player moving speed set to 3.73 unit/sec. It mimics the inside of a modern office building with rooms, corridors, and open areas of various sizes but lacks exterior windows. All doors inside the environment are “locked,” that is, participants are only allowed to navigate through the corridors and open areas. Eighty-two distinct landmarks, made up of 240 objects and 62 pictures, were placed at various locations (Figure 1). On the basis of criteria for significance (Lynch, 1960), the landmarks were divided into three categories. “Primary landmarks” have unambiguous shapes, standing in sharp contrast to the surroundings with a prominent position within the virtual environment. “Secondary landmarks” have unambiguous shapes and stand in sharp contrast to their surroundings, but with less prominent positions. “Minor landmarks” had an unambiguous shape but were only visible from a few locations and not that easy to separate from the surroundings. The main focus in the learning phase was on primary and secondary landmarks, and in the fMRI experiment, only primary and secondary landmarks were used as targets. This was done to ensure that the participants could learn the location of all possible target landmarks well enough to plan entire routes from start to target locations in the fMRI experiment. Wall structure, ceiling, carpeting, and lighting of the interior were similar throughout the VR environment and modeled to make the environment as realistic as possible.

Figure 1. 

Environmental tests. The top row shows an example of one of the Direction test trials (A) and an example of one of the Multiple distance test trials (B). The maps illustrate the types of representation required to complete each test type successfully. (A) In the Direction test, the participant determines the direction to the target landmark when facing the start landmark as shown in the figure. To achieve this, the participant has to first retrieve a fine-grained local representation, the exact position and orientation of the start landmark relative to the adjacent walls/architectural features in the background, and next where the target landmark was located relative to the start landmark, requiring only a coarse judgment of the landmark's locations. Two points were given for choosing the correct direction, and one point for choosing either of the two most nearby directions. (B) In the Multiple distance test, the participant has to remember which of four landmarks was farthest away from a target landmark. This required comparing interlandmark relationships in terms of Euclidian distances, in other words, a coarse, global representation of the environment. One point was given for the correct alternative.

Figure 1. 

Environmental tests. The top row shows an example of one of the Direction test trials (A) and an example of one of the Multiple distance test trials (B). The maps illustrate the types of representation required to complete each test type successfully. (A) In the Direction test, the participant determines the direction to the target landmark when facing the start landmark as shown in the figure. To achieve this, the participant has to first retrieve a fine-grained local representation, the exact position and orientation of the start landmark relative to the adjacent walls/architectural features in the background, and next where the target landmark was located relative to the start landmark, requiring only a coarse judgment of the landmark's locations. Two points were given for choosing the correct direction, and one point for choosing either of the two most nearby directions. (B) In the Multiple distance test, the participant has to remember which of four landmarks was farthest away from a target landmark. This required comparing interlandmark relationships in terms of Euclidian distances, in other words, a coarse, global representation of the environment. One point was given for the correct alternative.

Prescanning

Day 1

On the initial day of the experiment, the participants completed the learning phase. First, the participants freely explored the VR environment for 2 × 10 min using a standard desktop computer and a joystick. Next, they completed three test sequences to become familiarized with the VR in its full extent and to see all landmarks at least once. In each test sequence, the participants started at a specific landmark in the environment and had to find a target landmark that was shown in the bottom center of the screen. Upon arrival at the target landmark, a new target landmark was presented. The participants were given new target landmarks in the same manner until a sequence of approximately 25 landmarks was completed. In the first sequence primary landmarks were used as targets, in the second sequence secondary landmarks were used as targets, and in the third sequence minor landmarks were used. This was done to make sure the participants had seen as much as possible of the environment before they had to find the most difficult and/or unobtrusive landmarks. After completing all three sequences, the participants were tested on their proficiency in the VR by means of 18 navigation tasks. In each task, they had to find their way from a randomly selected position in the environment to one of the primary landmarks.

Day 2

On Day 2, the participants first explored the environment freely for 10 min. Next, they completed a test sequence where they had to find their way to 20 primary and 26 secondary landmarks. The primary and secondary landmarks were the same as the target landmarks in the fMRI experiment but presented in a different order. This was done to assure good performance during fMRI. Participants were excluded from the fMRI experiment if they made more than four errors of the 46 trials in the test sequence. Before fMRI the participants were given a 30-min break.

Scanning Procedure

Scanning was performed on a 3T Siemens Trio scanner with a 12-channel Head Matrix Coil (Siemens AG, Erlangen, Germany). Foam pads were used to minimize head motion. The fMRI stimuli were presented using MRI compatible LCD goggles with 800 × 600 resolution (Nordic Neuron Lab, Bergen, Norway), and the participants moved inside the environment using an MRI compatible joystick (Current Designs, Philadelphia, PA).

The participants were first allowed to familiarize themselves with the presentation equipment and joystick and then completed practice trials from the different experimental conditions. Scanning was commenced when complete task compliance was ensured.

fMRI Paradigm

The fMRI paradigm was a self-paced block design with alternating blocks of navigation (max duration = 40 ± 2 sec) and rest (cross fixation; 10 ± 2 sec).

Each navigation block consisted of three phases: Self-localization, Planning, and Execution. At the start of a navigation block, the participants were placed at a random position in the environment and instructed to figure out their current position using at least one other landmark in addition to the start landmark by turning and looking around. This phase is referred to as Self-localization. The participants were instructed to press a button when they were confident that they knew their current position. Subsequently, a target landmark was displayed in the bottom center of the screen. The participants were instructed to determine the location of this target landmark in the environment and plan as accurately as possible the path toward it. This phase is referred to as Planning. The participants pressed a button when they had finished planning and then started moving toward the target landmark. This phase is referred to as Execution. When the target was reached or the maximum allotted time had expired, the navigation block ended and the rest condition ensued.

Each participant completed three experimental runs, with 20 navigation blocks and 20 rest blocks in each run. The order of the runs was randomized between participants and the order of the blocks was randomized within each run.

Positional data capturing the participants' movements inside the environment were logged with a sampling interval of 30 msec.

Imaging Parameters

T2* weighted, BOLD-sensitive images were acquired during the navigation task using an EPI pulse sequence (repetition time = 2600 msec, echo time = 30 msec, field of view = 244 mm, slice thickness = 3.0 mm, slice number = 47, matrix = 80 × 80 giving an in-plane resolution of 3.0 × 3.0 mm). The slices were positioned as close to 90° on the anterior–posterior direction of the hippocampus as possible. Each functional run contained 449 ± 32 volumes, depending on the time needed by each individual to complete the runs. For anatomical reference, a T1-weighted 3-D volume was acquired with an MP-RAGE sequence (repetition time = 2300 msec, echo time = 30 msec, field of view = 256 mm, slice thickness = 1.0 mm, matrix 256 × 256 giving an in-plane resolution of 1.0 × 1.0 mm).

Postscanning Tests and Questionnaires

After scanning, the participants performed three computer-based tests to ascertain the level of fine-grained, local representations and coarse, global representations of the VR environment, inspired by Newcombe and Liben (1982). The first test, the Multiple distance test, was related to coarse, global knowledge of map distance in multiple cardinal directions between several distant landmarks in the virtual environment (see Figure 1). The participants had to decide which of four landmarks was farthest away from a target landmark. Knowledge of the distance in time when walking between the landmarks, for example, would not be sufficient to solve this problem, rather a coarse, global representation of the environment is needed. One point was given for the correct alternative. The second test, the Direction test, assessed the participants' ability to evaluate the direction between pairs of landmarks in the virtual environment. This test was divided into Local direction and Global direction. The Local direction test included pairs of nearby landmarks that were located within the same part of the environment, covering maximally one sixth of the total area. The Global direction test included pairs of distal landmarks that were located in two remote parts of the environment, stretching across at least two thirds of the total environment. Having a coarse, global representation of the environment would be especially advantageous for the Global direction test. Both the Local and Global direction tests first required evaluation of the exact position and orientation of the start landmark relative to the adjacent walls, that is, fine-grained, local information. Next, the start landmark had to be placed relative to a target landmark to be able to indicate the direction to the target landmark. The placement of the target landmark relative to the start landmark, compared with the local positioning of the start landmark, only requires a coarse representation of the environment. There were eight alternatives based on north–south and east–west directions. Two points were given for correct direction, and one was given for the two adjacent directions. In the third test, the Spatial sequence test, a more coarse, global environmental representation was assessed. The participants were shown four landmarks and had to put them in the correct order when pretending to move from one particular position to another, stretching across at least two thirds of the total environment. The sequences of landmarks were previously unexperienced and did not consist of landmarks along routes that were performed during environmental learning. Remembering the routes as previously experienced was thus insufficient to perform the task, which required information to be linked across the learned routes, representing a more coarse, global overview of the environment (Siegel & White, 1975). One point was given if all landmarks were placed in a correct sequence.

All participants completed a strategy questionnaire, which included 29 statements relating to the use of fine-grained, local environmental representations and a coarse, global representation of the environment during the Self-localization and Planning phases of navigation (see Table 1). The participants rated each statement on a 9-point scale, ranging from strongly agree (9) to strongly disagree (1).

Table 1. 

Strategy Questionnaire Results Describing the Average Group Score for the Mental Strategies Used in Navigational Self-localization and Planning



Mean + SD
Self-localization 
s1 A main landmark 7.0 ± 2.3 
s2 A map like representation of the environment 6.3 ± 2.3 
s3 A route or a sequence of landmarks 3.5 ± 2.0 
s4 Nearby landmarks 7.5 ± 1.5 
s5 Landmarks from the whole environment 4.8 ± 2.3 
s6 A representation of the environment as a whole 6.1 ± 2.3 
 
Planning 
p1 Placed the target landmark relative to nearby landmarks 6.6 ± 1.9 
p2 Placed the target landmark in a representation of the environment as a whole 6.8 ± 2.0 
p3 Placed the target landmark relative to start position 5.9 ± 2.5 
p4 Placed the target landmark relative to other landmarks in a route or sequence 4.1 ± 2.2 
p5 Direction to target landmark 8.3 ± 1.7 
p6 Direction to landmarks in between 4.5 ± 2.0 
p7 Distance to target landmark 5.4 ± 2.4 
p8 Distance to landmarks in between 3.0 ± 1.5 
p9 Sequences of landmarks 4.1 ± 1.9 
p10 Exact routes 3.6 ± 2.1 
p11 A map-like representation of the environment 6.0 ± 2.4 
p12 A representation of the environment as whole 6.0 ± 2.6 
p13 Landmarks in between 5.2 ± 2.3 
p14 Number of turns 3.4 ± 2.5 
p16 A set of routes 3.8 ± 2.2 
p17 Rapid route following 5.3 ± 3.2 
p18 Direction only? 8.5 ± 0.8 
p19 I knew the environment well enough to plan 7.3 ± 1.1 
p20 Mental route following over long distances 6.0 ± 2.3 
p21 Mental route following over short distances 5.8 ± 2.7 
p22 Direction(alone) over long distances 6.8 ± 2.1 
p23 Direction(alone) over short distances 5.4 ± 2.4 


Mean + SD
Self-localization 
s1 A main landmark 7.0 ± 2.3 
s2 A map like representation of the environment 6.3 ± 2.3 
s3 A route or a sequence of landmarks 3.5 ± 2.0 
s4 Nearby landmarks 7.5 ± 1.5 
s5 Landmarks from the whole environment 4.8 ± 2.3 
s6 A representation of the environment as a whole 6.1 ± 2.3 
 
Planning 
p1 Placed the target landmark relative to nearby landmarks 6.6 ± 1.9 
p2 Placed the target landmark in a representation of the environment as a whole 6.8 ± 2.0 
p3 Placed the target landmark relative to start position 5.9 ± 2.5 
p4 Placed the target landmark relative to other landmarks in a route or sequence 4.1 ± 2.2 
p5 Direction to target landmark 8.3 ± 1.7 
p6 Direction to landmarks in between 4.5 ± 2.0 
p7 Distance to target landmark 5.4 ± 2.4 
p8 Distance to landmarks in between 3.0 ± 1.5 
p9 Sequences of landmarks 4.1 ± 1.9 
p10 Exact routes 3.6 ± 2.1 
p11 A map-like representation of the environment 6.0 ± 2.4 
p12 A representation of the environment as whole 6.0 ± 2.6 
p13 Landmarks in between 5.2 ± 2.3 
p14 Number of turns 3.4 ± 2.5 
p16 A set of routes 3.8 ± 2.2 
p17 Rapid route following 5.3 ± 3.2 
p18 Direction only? 8.5 ± 0.8 
p19 I knew the environment well enough to plan 7.3 ± 1.1 
p20 Mental route following over long distances 6.0 ± 2.3 
p21 Mental route following over short distances 5.8 ± 2.7 
p22 Direction(alone) over long distances 6.8 ± 2.1 
p23 Direction(alone) over short distances 5.4 ± 2.4 

Each individual was asked to rate on a scale from 1 to 9 to what extent he used the different strategies during the Self-localization and Planning phases (SD, standard deviation).

Finally the participants were given a Map test. They were shown a 2-D representation of the environment that included only the outer walls and no interior architecture or landmarks and were asked to place 14 primary landmarks in correct predefined locations on this 2-D representation. The participants received one point for each landmark placed in the correct location.

In summary each individual's success in forming a coarse, global representation of the environment was tested with the Spatial sequence, Global direction, Multiple distance, and Map tests, whereas fine-grained, local representations were tested with the Local direction test. In addition, each individual's own experience of using a coarse, global representation of the environment and/or fine-grained, local environmental representation was assessed based on self-ratings of strategy use.

Data Analysis

Behavioral Data

Behavioral data were analyzed in SPSS 17.0 (SPSS, Inc., Chicago, IL). The critical performance measures include duration and distance moved in the three navigation phases of Self-localization, Planning, and Execution, the average success rate for the navigation tasks during fMRI, and the average success rate on the postscan tests of Multiple distance, Local direction, Global direction, and Spatial sequence. The strategies participants used during navigation were inferred from the strategy questionnaire data. Here, a principal component analysis (PCA) was used to extract the main factors describing the reported strategies, separately for the questions related to Self-localization and Planning. Factors were made orthogonal using varimax rotation with Kaiser Normalization, and questions with an absolute coefficient value of less than 0.4 for a given factor were suppressed. We used factor analysis instead of simply aggregating scores regarding fine-grained environmental representations and a coarse environmental representation separately, because factor analysis is a more objective and data-driven approach. Although factor analysis is usually considered to be a technique for relatively large samples, it has been shown that sample sizes far below 50 can give reliable results (De Winter, Dodou, & Wieringa, 2009). To confirm this, correlation analyses between questionnaire scores and fMRI activity were performed with the score from the individual questions as well as with the calculated factors. Similar results were observed. The scores from the environmental tests were tested for correlation, between participants, using Spearman's rho.

MRI Data Analysis

Imaging data were analyzed using FSL 4.1.8 (Analysis Group, FMRIB, Oxford, UK). First, nonbrain tissue was removed from the T1 anatomical images, and the resulting images were transformed to the MNI standard template (1 × 1 × 1 mm; Montreal Neurological Institute, Montreal, QC, Canada) using nonlinear registration (FNIRT; reference). The fMRI data were motion corrected, smoothed with a 5-mm FWHM Gaussian filter, and temporally high-pass filtered with a cutoff at 250 sec. Each functional image series was coregistered to the corresponding anatomical T1 image and transformed into MNI space using the transformation matrix obtained with the T1 image. The statistical analysis of the fMRI data was carried out in FEAT (Smith et al., 2004). The experimental conditions (Self-localization, Planning, and Execution) were modeled according to a boxcar stimulus function convolved with a two-gamma hemodynamic response function (Boynton, Engel, Glover, & Heeger, 1996). The rest condition constituted the model's baseline. The effect of each condition was estimated with a general linear model (GLM) and averaged across participants in a mixed effects analysis (FLAME 1; Beckmann, Jenkinson, & Smith, 2003).

Medial-temporal Lobe Analysis

A brain mask was applied to investigate activation in the medial-temporal lobe (MTL) more closely. The mask was created from the probabilistic maps of the Harvard Oxford Structural Atlases (part of FSL; fsl.fmrib.ox.ac.uk/fsl/fslwiki/Atlases), using no probability threshold. The mask contained 76,864 (1-mm) voxels and included both the hippocampus and parahippocampal gyrus. Within this mask, specific ROIs were also identified. The entorhinal cortex and the perirhinal cortex were defined based on anatomical boundaries (Insausti et al., 1998). The hippocampus was divided into head, body, and tail (DeFelipe et al., 2007; Duvernoy, 2005).

The contrasts Self-localization > Execution and Planning > Execution were investigated within participants using paired t tests together with an initial voxel threshold of p < .001 and a subsequent clusterwise correction for multiple comparisons (p < .05).

Combined fMRI and Behavioral Data Analysis

We performed separate GLM analyses to examine possible correlations between activation in the MTL ROI and individual scores on the environmental tests and strategy questionnaire. Each individual's results on the Multiple distance, Local direction, Global direction, and Spatial sequence tests were entered as separate regressors in a mixed effects GLM analysis with MTL activation in the contrasts Planning > Execution and self-localization > Execution as the dependent factors. The threshold was set at voxel p = .005 (uncorrected), and a minimum cluster size of 30 continuous 1-mm voxels.

Each individual's strategy questionnaire factor loadings (see above for details on factor analysis) were entered as separate regressors in two GLM analyses, one with MTL activation in the contrast Self-localization > Execution as the dependent factor and one with MTL activation in the contrast Planning > Execution as the dependent factor. In the analysis of Self-localization > Execution, the factor loadings relating to Self-localization were entered (s1–6; Table 2). In the analysis of Planning > Execution, the factor loadings relating to Planning were entered (p1–23; Table 3).

Table 2. 

Factor Analysis of Questionnaire Items Related to Self-localization



Local Environment Factor S1
Environmental Coherence Factor S2
Route/Sequence Factor S3
s4 Nearby landmarks 0.873   
s2 A map like representation of the environment −0.774   
s6 A representation of the environment as a whole −0.467 0.443 −0.433 
s5 Landmarks from the whole environment  0.781  
s1 A main landmark  −0.735  
s3 A route or a sequence of landmarks   0.89 


Local Environment Factor S1
Environmental Coherence Factor S2
Route/Sequence Factor S3
s4 Nearby landmarks 0.873   
s2 A map like representation of the environment −0.774   
s6 A representation of the environment as a whole −0.467 0.443 −0.433 
s5 Landmarks from the whole environment  0.781  
s1 A main landmark  −0.735  
s3 A route or a sequence of landmarks   0.89 

The factors were extracted from the strategy questionnaire scores using a PCA and then rotated using varimax. Questions from the strategy questionnaire with an absolute coefficient value of less than 0.4, for a given factor, were suppressed. Factor S1 was given the name “Local Environment” because it represents the use of nearby landmarks, but not a map like representation of the environment or a representation of the environment as a whole. Factor S2, representing the use of a coherent representation of the environment as a whole, and not just an independent main landmark, and was given the name “Environmental Coherence.” Factor S3 was named “Route/sequence,” representing use of a route or a sequence of landmarks, but not the environment as a whole.

Table 3. 

Factor Analysis of Questionnaire Items Related to Planning



Environmental Coherence Factor P1
Sequence and Turns Factor P2
Direction Factor P3
Exact/Complete Long Routes Factor P4
Environmental Chunking Factor P5
Map Factor P6
Direction to Remote Landmarks Factor P7
Routes Factor P8
p12 A representation of the environment as whole 0.786        
p7 Distance to target landmark 0.784        
p2 In a representation of the environment as a whole 0.774        
p14 The number of turns  0.734       
p4 Relative to other landmarks in route or sequence  0.707       
p23 Direction(alone) over short distances  −0.638       
p9 Sequences of landmarks  0.613    −0.423   
p6 Direction to landmarks in between  0.591       
p5 Direction to target landmark   0.773      
p18 Direction   0.769      
p17 Rapid route following 0.47  0.723      
p20 Mental route following over long distances    0.835     
p10 Exact routes    0.811     
p1 Relative to nearby landmarks   −0.43 −0.527     
p8 Distance to landmarks in between     0.833    
p16 A set of routes     −0.687   0.428 
p21 Mental route following over short distances     0.629    
p11 A map like representation(s) of the environment      0.794   
p13 Landmarks in between     0.408 −0.633   
p22 Direction(alone) over long distances       0.762  
p19 I knew the environment well enough to plan       0.725  
p3 Relative to start position        −0.915 


Environmental Coherence Factor P1
Sequence and Turns Factor P2
Direction Factor P3
Exact/Complete Long Routes Factor P4
Environmental Chunking Factor P5
Map Factor P6
Direction to Remote Landmarks Factor P7
Routes Factor P8
p12 A representation of the environment as whole 0.786        
p7 Distance to target landmark 0.784        
p2 In a representation of the environment as a whole 0.774        
p14 The number of turns  0.734       
p4 Relative to other landmarks in route or sequence  0.707       
p23 Direction(alone) over short distances  −0.638       
p9 Sequences of landmarks  0.613    −0.423   
p6 Direction to landmarks in between  0.591       
p5 Direction to target landmark   0.773      
p18 Direction   0.769      
p17 Rapid route following 0.47  0.723      
p20 Mental route following over long distances    0.835     
p10 Exact routes    0.811     
p1 Relative to nearby landmarks   −0.43 −0.527     
p8 Distance to landmarks in between     0.833    
p16 A set of routes     −0.687   0.428 
p21 Mental route following over short distances     0.629    
p11 A map like representation(s) of the environment      0.794   
p13 Landmarks in between     0.408 −0.633   
p22 Direction(alone) over long distances       0.762  
p19 I knew the environment well enough to plan       0.725  
p3 Relative to start position        −0.915 

The factors were extracted from the strategy questionnaire scores using a PCA and then rotated using varimax. Questions from the strategy questionnaire with an absolute coefficient value of less than 0.4 for a given factor were suppressed. Factor P1 was named “Environmental Coherence” because it involves primarily questions indicating the use of a coherent representation of the environment. Factor P2 was given the name “Sequences and turns,” because it involves questions that indicate the use of landmark sequences and turns and direction to landmarks in between but not direction, for example, to the target landmark. Factor P3 was named “Direction,” the name is based on the fact that two of the four questions in this factor depend on the use of direction. Factor P4, was given the name “Exact/Complete Long Routes,” representing the use of exact routes and mental route following over long distances. Factor P5 was named “Environmental chunking,” because the questions involved representation of a limited part of the environment between the start position and the target landmark. Factor P6 was given the name “Map,” because it involves using a map-like representation(s) in a third person perspective, but not parts of the environment like sequences of landmarks. Factor P7 was named “Direction to remote landmarks” because it involves using direction between the start position and remote landmarks. Factor P8 was named “Routes,” since it involves using a set of routes when planning, but not placing the target landmark relative to the start position.

We performed an additional GLM analysis to investigate possible associations between activation in a parietal ROI (see below for further details) and different types of route representations. Here, we used all questionnaire items related to the use of route centered representations during Planning: p9 (landmark sequences), p10 (exact routes), p14 (nr of turns), and p16 (set of routes). Each individual's item ratings were entered as separate repressors in the GLM. The contrast used was Planning > Execution. The statistical threshold was voxel p < .005 (uncorrected) with a minimum cluster size of 30 continuous 1-mm voxels. A stricter statistical threshold was used for the parietal lobe, compared with the MTL, since the calculated t and β values are reduced for subcortical compared with cortical structures using a Siemens 3T system with a 12-channel head coil (Kaza, Klose, & Lotze, 2011). The parietal ROI consisted of 288,809 (1-mm) voxels that encompassed the parietal lobe. The mask was based on the probabilistic maps of the MNI structural atlas (part of FSL; fsl.fmrib.ox.ac.uk/fsl/fslwiki/Atlases), using a probability threshold of 25%.

Hippocampal Shape Analysis

The automatic segmentation tool within FSL (FIRST) was applied to the T1 images to estimate individual variations in shape and size of the right and left hippocampus. Each segmented hippocampus was transformed to a template based on the averages of all participants' common mean native space, using six degrees of freedom, thus only differences in local shape or local volumes were assessed. To investigate the use of a coarse, global representation of the environment independent of navigational phase, the questions from the strategy questionnaire most strongly related to the use of a global environmental representation for both Self-localization and Planning (s2, s5, s6, p2, p11, p12) were grouped, and an aggregate score was calculated. Likewise, the questions from the strategy questionnaire most strongly related to using fine-grained, local environmental representations for both self-localization and planning (s4, p5, p14, p18) were grouped, and an aggregate score was calculated. These scores, plus scores from the Multiple distance and Global and Local direction tests, were entered as separate regressors in a GLM. This was done to investigate the relationship between hippocampal shape and the type of environmental representation acquired and applied during navigation, as well as the level of spatial proficiency as determined by the different tests in each participant. We used a statistical threshold of p = .0001, uncorrected.

RESULTS

Learning Phase

Two participants were excluded from the analysis: one because he failed to meet the learning criterion and the other because of severe nausea during VR navigation in the scanner. After the learning phase on Day 1, the participants included in the final sample (n = 28) were able to reach 88.9 ± 7.4% of the landmarks in the environment in the prescan test, indicating they were highly familiar with the environment.

fMRI Task Performance

The Self-localization phase lasted on average 8.5 ± 2.9 sec, the Planning phase lasted 8.1 ± 2.5 sec, and the Execution phase lasted 13.2 ± 1.0 sec. There was a significant effect of phase with regard to duration (F = 47.5, p < .001). Post hoc comparisons revealed a significant difference between Self-localization and Execution (t = 4.7, p < .001) and between Planning and Execution (t = 5.1, p < .001), but not between Self-localization and Planning. Participants moved on average 2.7 ± 2.7 m during Self-localization, 0.27 ± 0.34 m for Planning, and 100.2 ± 5.71 m during Execution. There was a significant effect of phase with regard to distance moved (F = 5480, p < .001). Post hoc comparisons showed a significant difference between Self-localization and Planning (t = 2.4, p < .0001) and, as expected, between Self-localization and Execution (t = 95.8, p < .0001), and between Planning and Execution (t = 97.8, p < .0001). Movement plots for each trial clearly showed that the participants followed the instructions for movement in the different trial phases, that is, moving around the starting landmark during Self-localization and standing still while choosing the direction or route during Planning.

The participants were able to reach on average 82.2 ± 14.9% of the target landmarks during scanning. Visual inspection of the movement plots, combined with feedback from the participants after scanning, revealed that in most cases participants failed to reach the target because they ran out of time, and not because they did not know the way to the target landmark.

Tests of VR Knowledge and Navigation Strategies

The average number of correct answers were as follows: on the Multiple distance test, 8.1 ± 1.7 of 10 points; on the Local direction test, 7.8 ± 1.6 of 10 points; on the Global direction test, 7.9 ± 1.9 of 10 points; on the Spatial sequence test, 5.4 ± 2.3 of 10 points; and on the Map test, 13.6 ± 1 of 14 points. The average success rate was well above chance level for all tests, which was 2.4 point for the Multiple distance test, 2.5 points for the Local direction and Global direction test, 2.5 points for the Spatial sequence test, and 1 point for the Map test. These results show that the participants knew the VR environment very well, although they did not reach all target landmarks during fMRI within the given time. The scores on the Local direction test and Global direction test were not correlated (R = .163, p = .407), nor were the scores on the Local direction test and Multiple distance test (R = .294, p = .129), the scores on the Global direction test and Multiple distance test (R = .242, p = .215), the scores on the Local direction test and the Spatial sequence test (R = .257, p = .186), or the scores on the Global direction test and the Spatial sequence test (R = .315, p = .102). However, a significant correlation was found between the score on the Multiple distance test and the score on the Spatial sequence test (R = .573, p = .001). These results support a behavioral, and possibly also neural, separation of the ability to have a highly accurate spatial (environmental) representation of individual landmarks versus a more coarse, global environmental representation.

On the basis of the factor analysis of the questionnaire, the mental representations used for Self-localization (S) could be linked to the Environmental coherence factor S2, which describes the use of a coherent representation of the environment; the Local environment factor S1, which describes the use of nearby landmarks, and the Route/sequence factor S3, which depends on positioning one's self in a more local route or sequence of landmarks (Table 2).

Planning (P) relied on eight factors, of which half can be viewed as reflecting the use of a coarse, global environmental representation and the other half the use of fine-grained, local environmental representations. The coarse, global factors included Environmental coherence (P1), which describes the use of a coherent representation of the entire environment; Exact long routes (P4), which described the use of an exact representation of the entire route to distant targets; Map-like representation (P6), which indicated the use of a mental map of the environment; and Direction (P7), which described the use of direction between start position and remote landmarks other than the target landmark. The factors that represented fine-grained, local environmental representations were Sequences and turns (P2), which depend on remembering and utilizing landmark sequences and number of turns to reach target; Local direction (P3), which describes using the direction from start position to target; Exact routes (P4), which indicates using an exact representation of the entire route from start position to target; Environmental chunking (P5), which describes chunking part(s) of the environment, including landmarks located between start position and target, and uses these halfway targets to find the final target; and Route (P8), which describes using a set of fixed routes. Note that the factor Exact routes (P4) is linked to the use of both a global environmental representation and fine-grained, local representations. This was because of the fact that both of the questions “Mental route following over long distances” and “Use Exact routes” load on this factor (Table 3).

fMRI Results

MTL Activation during Different Phases of Navigation

For the contrasts Self-localization > Execution and Planning > Execution, increased activation was present in the right hippocampal body, bilateral hippocampal head, and bilateral rostral entorhinal cortex (Table 4). For Self-localization > Execution, additional activations were observed in the left hippocampal body, right caudal entorhinal cortex, and bilateral parahippocampal cortex. Only the contrast Planning > Execution activated the hippocampal tail bilaterally.

Table 4. 

MTL Activation during Self-localization and Planning

MTL Region
Coordinates of Peak Activation (MNI)Cluster Number
Cluster Size (No. of Voxels)
Z Score
x
y
z
Self-localization > Execution 
Hippocampal body 25 −30 −10 (1)  3.56 
−16 −31 −7 (2)  4.02 
Hippocampal head 28 −16 −25 (1)  4.09 
−30 −18 −22 (2)  3.49 
Entorhinal cortex (caudal) 25 −22 −24 (1)  4.03 
Entorhinal cortex (rostral) 22 −5 −37 (1)  4.24 
Perirhinal cortex 25 −36 (1)  4.68 
−30 −8 −37 (2)  4.01 
Parahippocampal cortex 20 −32 −23 (1)  4.41 
−27 −29 −25 (2)  3.67 
Temporal fusiform cortex 39 −10 −39 13281 5.16 
Temporal pole −31 −38 7101 5.53 
 
Planning > Execution 
Hippocampal tail 13 −38 (1)  4.74 
−14 −40 (2)  4.48 
Hippocampal body −28 −27 −17 (2)  3.97 
Hippocampal head 26 −16 −21 (1)  3.54 
−31 −11 −24 (2)  4.01 
Entorhinal cortex (rostral) 17 −28 (1)  4.66 
−21 −30 (2)  4.52 
−25 −11 −34 (2)  3.69 
Perirhinal cortex 32 −4 −34 (1)  4.46 
Temporal fusiform cortex 39 −11 −38 9900 5.08 
Temporal pole −29 −38 11473 4.88 
MTL Region
Coordinates of Peak Activation (MNI)Cluster Number
Cluster Size (No. of Voxels)
Z Score
x
y
z
Self-localization > Execution 
Hippocampal body 25 −30 −10 (1)  3.56 
−16 −31 −7 (2)  4.02 
Hippocampal head 28 −16 −25 (1)  4.09 
−30 −18 −22 (2)  3.49 
Entorhinal cortex (caudal) 25 −22 −24 (1)  4.03 
Entorhinal cortex (rostral) 22 −5 −37 (1)  4.24 
Perirhinal cortex 25 −36 (1)  4.68 
−30 −8 −37 (2)  4.01 
Parahippocampal cortex 20 −32 −23 (1)  4.41 
−27 −29 −25 (2)  3.67 
Temporal fusiform cortex 39 −10 −39 13281 5.16 
Temporal pole −31 −38 7101 5.53 
 
Planning > Execution 
Hippocampal tail 13 −38 (1)  4.74 
−14 −40 (2)  4.48 
Hippocampal body −28 −27 −17 (2)  3.97 
Hippocampal head 26 −16 −21 (1)  3.54 
−31 −11 −24 (2)  4.01 
Entorhinal cortex (rostral) 17 −28 (1)  4.66 
−21 −30 (2)  4.52 
−25 −11 −34 (2)  3.69 
Perirhinal cortex 32 −4 −34 (1)  4.46 
Temporal fusiform cortex 39 −11 −38 9900 5.08 
Temporal pole −29 −38 11473 4.88 

The analysis was carried out using an MTL mask and voxel based thresholding, p = .001 uncorrected, and a cluster threshold, p = .05 corrected. MNI 152 brain template has a voxel resolution of 1 mm3. The cluster number is given in parenthesis for secondary peaks within the respective clusters. R = right; L = left.

The Association between MTL Activity and Multiple Distance and Direction Tests Scores

There was no association between the spatial test scores and activation in the contrast Self-localization > Execution. For Planning > Execution, there was an effect in the hippocampal tail for both Local direction and Global direction, whereas in the hippocampal head an effect was observed only for Global direction (Figure 2; Table 5). The Multiple distance test showed an effect only in the right hippocampal head. No effects were observed of the Spatial sequence test.

Figure 2. 

Fine-grained, local representations in posterior hippocampus and a coarse, global environmental representation anteriorly. Regions in MTL that correlated with scores on Local direction and Multiple distance test (A) and strategy factors (B), for either Self-localization > Execution or Planning > Execution. (A1) shows areas correlating with the Local direction test, which tested the ability to evaluate the exact position of the landmark within the proximate surrounding walls to be able to tell the direction to a nearby landmark. The images in the left column, B1–B3, represent analysis of MTL activity related to more exact knowledge of the position of the landmarks within the environment. B1–B3 show areas positively correlated with the strategy factors from the factor analysis of the strategy questionnaire, representing the use of details from a more local part of the environment. The images in the right column show MTL regions that correlated with the Multiple distance test (A2), which tested if the participants were able to decide the relative distance between several remote landmarks in the environment and (B4–B6) show MTL regions that correlated with strategy factors, indicating the use of a coarse, global representation of the environment. Voxel-based thresholding was applied at the level of p = .05 uncorrected for the factor GLM and p = .005 uncorrected for test score GLM. These images are all from the right hemisphere. Activations are superimposed on the MNI 152 brain template.

Figure 2. 

Fine-grained, local representations in posterior hippocampus and a coarse, global environmental representation anteriorly. Regions in MTL that correlated with scores on Local direction and Multiple distance test (A) and strategy factors (B), for either Self-localization > Execution or Planning > Execution. (A1) shows areas correlating with the Local direction test, which tested the ability to evaluate the exact position of the landmark within the proximate surrounding walls to be able to tell the direction to a nearby landmark. The images in the left column, B1–B3, represent analysis of MTL activity related to more exact knowledge of the position of the landmarks within the environment. B1–B3 show areas positively correlated with the strategy factors from the factor analysis of the strategy questionnaire, representing the use of details from a more local part of the environment. The images in the right column show MTL regions that correlated with the Multiple distance test (A2), which tested if the participants were able to decide the relative distance between several remote landmarks in the environment and (B4–B6) show MTL regions that correlated with strategy factors, indicating the use of a coarse, global representation of the environment. Voxel-based thresholding was applied at the level of p = .05 uncorrected for the factor GLM and p = .005 uncorrected for test score GLM. These images are all from the right hemisphere. Activations are superimposed on the MNI 152 brain template.

Table 5. 

The Relationship between the Spatial Accuracy and Multiple Distance Test Scores and MTL Activation during Planning

MTL Region
Coordinates of Peak Activation (MNI)
Cluster Number
Cluster Size (No. of Voxels)
Z Score
x
y
z
Global Direction Test 
Hippocampal tail 23 −42 5399 4.58 
−25 −41 2588 3.53 
Hippocampal head 27 −15 −15 141 3.10 
−25 −14 −15 (1)  3.41 
Parahippocampal cortex 28 −34 −18 (1)  3.06 
Entorhinal cortex, caudal −20 −17 −31 2559 3.95 
28 −21 −27 (5)  3.15 
Perirhinal cortex 11 −11 −23 1899 3.61 
24 −33 186 3.3 
−29 −30 1306 3.63 
−35 −16 −36 167 3.48 
 
Local Direction Test 
Hippocampal tail 12 −38 603 3.51 
 
Multiple Distance Test 
Hippocampal head 26 −16 −17 35 2.78 
 
Spatial Sequence Test 
n/a       
MTL Region
Coordinates of Peak Activation (MNI)
Cluster Number
Cluster Size (No. of Voxels)
Z Score
x
y
z
Global Direction Test 
Hippocampal tail 23 −42 5399 4.58 
−25 −41 2588 3.53 
Hippocampal head 27 −15 −15 141 3.10 
−25 −14 −15 (1)  3.41 
Parahippocampal cortex 28 −34 −18 (1)  3.06 
Entorhinal cortex, caudal −20 −17 −31 2559 3.95 
28 −21 −27 (5)  3.15 
Perirhinal cortex 11 −11 −23 1899 3.61 
24 −33 186 3.3 
−29 −30 1306 3.63 
−35 −16 −36 167 3.48 
 
Local Direction Test 
Hippocampal tail 12 −38 603 3.51 
 
Multiple Distance Test 
Hippocampal head 26 −16 −17 35 2.78 
 
Spatial Sequence Test 
n/a       

For an explanation of the tests, see Methods. The analysis was carried out using an MTL mask and voxel based thresholding, p = .005 uncorrected. Only clusters with a cluster size of >30 voxels were reported. MNI 152 brain template has a voxel resolution of 1 mm3. The cluster number is given in parenthesis for secondary peaks within the respective clusters. R = right; L = left.

The Association between MTL Activity and Self-reported Navigation Strategy

The GLM revealed no significant relationship between activation in the MTL and any of the questionnaire factors when using the a priori statistical threshold. However, we did identify a number of sub-threshold activations (p < .05 uncorrected). Activation bilaterally in the hippocampal head correlated with all the factors related to the use of a coarse, global representation of the environment during both Self-localization and Planning. The factor Environmental coherence (S2) correlated with activation in the hippocampal head for the contrast Self-localization > Execution (Figure 2; Table 6), and the factors Exact route (P4), Map (P6), Direction to remote landmarks (P7), and Direction (P3) correlated with activation in the hippocampal head for the contrast Planning > Execution (Figure 2; Table 7). Hippocampal tail activation correlated with factors related to fine grained local environmental representations: Sequence and turns (P2), Exact routes (P4), and Direction (P3) (Figure 2; Table 6) for Planning > Execution.

Table 6. 

The Relationship between Factors of the Strategy Questionnaire and MTL Activation during Self-localization

MTL Region
Coordinates of Peak Activation (MNI)
Cluster Number
Cluster Size (No. of Voxels)
Z Score
x
y
z
Local (S1) 
n/a       
 
Environmental Coherence (S2) 
Hippocampal head −30 −15 −18 47 2.05 
24 −14 −17 39 2.05 
Perirhinal cortex −31 −39 823 3.54 
24 10 −26 219 2.74 
 
Route (S3) 
n/a       
MTL Region
Coordinates of Peak Activation (MNI)
Cluster Number
Cluster Size (No. of Voxels)
Z Score
x
y
z
Local (S1) 
n/a       
 
Environmental Coherence (S2) 
Hippocampal head −30 −15 −18 47 2.05 
24 −14 −17 39 2.05 
Perirhinal cortex −31 −39 823 3.54 
24 10 −26 219 2.74 
 
Route (S3) 
n/a       

The analysis was carried out using an MTL mask and voxel based thresholding, p = .05 uncorrected. Only clusters with a cluster size of >30 voxels were reported. MNI 152 brain template has a voxel resolution of 1 mm3. R = right; L = left.

Table 7. 

The Relationship between Factors of the Strategy Questionnaire and MTL Activation during Planning

MTL Region
Coordinates of Peak Activation (MNI)
Cluster Number
Cluster Size (No. of Voxels)
Z Score
x
y
z
Environmental Coherence (P1) 
n/a       
Sequence and Turns (P2) 
Hippocampal tail 33 −42 −1 744 2.38 
31 −38 (1)  2.33 
Hippocampal body −31 −32 −5 43 1.94 
Entorhinal cortex, caudal −14 −25 −19 562 2.32 
Perirhinal cortex −32 −4 −28 280 2.39 
 
Direction (P3) 
Hippocampal tail 16 −42 794 2.64 
22 −35 (1)  2.04 
Hippocampal body −33 −27 −7 1552 2.89 
Hippocampal head 29 −14 −14 210 1.94 
−26 −16 −10 (2)  2.58 
Parahippocampal cortex 34 −35 −15 192 2.09 
Perirhinal cortex −29 −28 308 2.28 
 
Exact Long Routes (P4) 
Hippocampal tail −12 −38 1754 2.83 
14 −35 1189 2.60 
Hippocampal head −31 −7 −22 (3)  2.52 
Entorhinal cortex, rostral 18 −27 112 2.32 
−24 −14 −37 46 2.25 
Perirhinal cortex −35 −18 −33 3138 2.76 
33 −24 −28 381 2.21 
28 −39 378 2.36 
−27 −39 49 1.94 
 
Environmental Chunking (P5) 
n/a       
 
Map (P6) 
Hippocampal head 25 −16 −8 241 2.27 
30 −19 −18 47 1.81 
 
Direction to Remote Landmarks (P7) 
Hippocampal body 39 −35 −7 430 2.38 
20 −24 −14 87 1.91 
15 −34 −8 46 1.87 
16 −28 −11 32 1.83 
Hippocampal head 18 −12 −15 314 2.51 
−33 −19 −12 266 1.96 
Entorhinal cortex, rostral 18 −8 −37 1626 3.07 
Amygdala 24 −2 −24 73 1.92 
 
Routes (P8) 
n/a       
MTL Region
Coordinates of Peak Activation (MNI)
Cluster Number
Cluster Size (No. of Voxels)
Z Score
x
y
z
Environmental Coherence (P1) 
n/a       
Sequence and Turns (P2) 
Hippocampal tail 33 −42 −1 744 2.38 
31 −38 (1)  2.33 
Hippocampal body −31 −32 −5 43 1.94 
Entorhinal cortex, caudal −14 −25 −19 562 2.32 
Perirhinal cortex −32 −4 −28 280 2.39 
 
Direction (P3) 
Hippocampal tail 16 −42 794 2.64 
22 −35 (1)  2.04 
Hippocampal body −33 −27 −7 1552 2.89 
Hippocampal head 29 −14 −14 210 1.94 
−26 −16 −10 (2)  2.58 
Parahippocampal cortex 34 −35 −15 192 2.09 
Perirhinal cortex −29 −28 308 2.28 
 
Exact Long Routes (P4) 
Hippocampal tail −12 −38 1754 2.83 
14 −35 1189 2.60 
Hippocampal head −31 −7 −22 (3)  2.52 
Entorhinal cortex, rostral 18 −27 112 2.32 
−24 −14 −37 46 2.25 
Perirhinal cortex −35 −18 −33 3138 2.76 
33 −24 −28 381 2.21 
28 −39 378 2.36 
−27 −39 49 1.94 
 
Environmental Chunking (P5) 
n/a       
 
Map (P6) 
Hippocampal head 25 −16 −8 241 2.27 
30 −19 −18 47 1.81 
 
Direction to Remote Landmarks (P7) 
Hippocampal body 39 −35 −7 430 2.38 
20 −24 −14 87 1.91 
15 −34 −8 46 1.87 
16 −28 −11 32 1.83 
Hippocampal head 18 −12 −15 314 2.51 
−33 −19 −12 266 1.96 
Entorhinal cortex, rostral 18 −8 −37 1626 3.07 
Amygdala 24 −2 −24 73 1.92 
 
Routes (P8) 
n/a       

The analysis was carried out using an MTL mask and voxel based thresholding, p = .05 uncorrected. Only clusters with a cluster size of >30 voxels were reported. MNI 152 brain template has a voxel resolution of 1 mm3. R = right; L = left.

The Association between Parietal Cortex Activity and Type of Mental Route Representations

A positive relationship was observed between the questionnaire items that indicate use of different types of route representations and activity in the inferior parietal cortex in the contrast Planning > Execution (Figure 3). More complete mental route representations were observed in anterior compared with posterior inferior parietal cortex. Most posteriorly, at the parietal-occipital junction, activation in left hemisphere was associated with the use of number of turns ([−34 −86 36], z = 3.52). More anteriorly, the use of sequences of landmarks was associated with activity in the right inferior parietal cortex ([41 −66 28], z = 3.47) and the use of complete routes with activity even more anteriorly in left inferior parietal cortex ([−52 −53 42], z = 5.18).

Figure 3. 

The route representation becomes more complete in the anterior part of the inferior parietal lobe. This figure shows the peak voxels in the inferior parietal lobe when correlating activation for condition Planning > Execution with route related questions from the strategy questionnaire. The original x coordinate of the peak voxels is presented as located to one hemisphere to visualize the relative position of each peak voxel along the anterior–posterior axis. Activation peaks are superimposed on the MNI 152 brain template.

Figure 3. 

The route representation becomes more complete in the anterior part of the inferior parietal lobe. This figure shows the peak voxels in the inferior parietal lobe when correlating activation for condition Planning > Execution with route related questions from the strategy questionnaire. The original x coordinate of the peak voxels is presented as located to one hemisphere to visualize the relative position of each peak voxel along the anterior–posterior axis. Activation peaks are superimposed on the MNI 152 brain template.

Hippocampal Shape Analysis

In the right hippocampal head, larger local volume correlated with self-reported use of a coarse, global representation of the environment, both during Self-localization and Planning (Figure 4). In the right hippocampal tail, smaller local volume was associated with the same measure (Figure 4). For the left hippocampus, there was no relationship between local volume and spatial representation. There were no significant relationships between local hippocampal volume and any of the test scores.

Figure 4. 

The relationship between hippocampal shape and the ability to use a coarse, global representation of the environment from an aggregate score, based on the strategy questionnaire (s5, s6, p2, and p12). Arrows pointing outward indicate increased local volume (enlargement), and arrows pointing inward indicate reduced local volume (reduced volume). Right hippocampal tail volume correlated negatively and right hippocampal head volume positively with this aggregated coarse, global score. To investigate local changes within the hippocampus, each segmented hippocampus was transformed to a common mean native space, using six degrees of freedom, whereas global changes from each hippocampus were removed. Threshold used was p = .0001 uncorrected. The color bar illustrates what z value the colors in the volumetric image represent.

Figure 4. 

The relationship between hippocampal shape and the ability to use a coarse, global representation of the environment from an aggregate score, based on the strategy questionnaire (s5, s6, p2, and p12). Arrows pointing outward indicate increased local volume (enlargement), and arrows pointing inward indicate reduced local volume (reduced volume). Right hippocampal tail volume correlated negatively and right hippocampal head volume positively with this aggregated coarse, global score. To investigate local changes within the hippocampus, each segmented hippocampus was transformed to a common mean native space, using six degrees of freedom, whereas global changes from each hippocampus were removed. Threshold used was p = .0001 uncorrected. The color bar illustrates what z value the colors in the volumetric image represent.

DISCUSSION

In this study, we tested the hypothesis that anterior hippocampus supports a coarse, global environmental representation whereas posterior hippocampus supports fine-grained, local representations. We used a virtual environment combined with fMRI to describe how localization of functional activation and hippocampal shape related to the level of coarse, global and fine-grained, local spatial representations of the virtual environment, measured with both objective tests and self-report. Our main finding was that activation in the hippocampal head was positively related to measures reflecting coarse, global environmental representations, whereas activation in the hippocampal tail was positively related to measures reflecting fine-grained, local representations. This is in line with the suggested functional segregation along the anterior–posterior axis of the hippocampus from animal studies of hippocampal function (Kjelstrup et al., 2008; Jung et al., 1994), but to our knowledge, this hypothesis has never before been tested in humans.

The behavioral data support the idea of two systems of spatial representation, one related to a coarse and global overview and another related to fine-grained, local environmental representations. The strategies that participants used during self-localization and navigational planning were reduced to 11 main factors that reflect the use of coarse, global environmental representations or fine-grained, local environmental representations. An example of a coarse, global environmental factor is Map, reflecting the use of an environmental map of the environment when planning how to reach the target. An example of a fine-grained, local factor is Sequence and turns, which involved evaluating landmarks and turns between the start position and the target landmark when planning how to reach the target. The idea of two spatial systems is supported by the observation that individual scores on the Local direction test were uncorrelated with the test scores that reflect processing of a coarse, global environmental representation, that is, Global direction, Multiple distance, and Spatial sequence test scores. This is consistent with a previous navigation study, where the representations of individual rooms were more accurate than and separate from a coarse, global representation of the environment (Colle & Reid, 2000). Although two separable spatial systems appear to exist, they are likely to interact. For example, people have been shown to switch between the two types of representations during navigation (Wang & Brockmole, 2003). We identified one factor, Exact long routes, that seems to reflect the use of both fine-grained, local environmental representations and a coarse, global environmental representation, also suggesting that the two systems interact.

The fMRI data show that representation of a coarse, global representation relies most strongly on the hippocampal head. The Multiple distance test score was only related to activation in the hippocampal head. Importantly, activation in the hippocampal head was associated with the Global direction test, but not the Local direction test. Moreover, activation in the hippocampal head was related to the questionnaire factor Environmental coherence during Self-localization and to the factors Map and Exact long route during Planning. These factors are all considered to reflect the use of a coarse, global representation of the environment. The activation in the hippocampal head did not correlate with the Spatial sequence test. This might be because of low performance on this test, or that this test is not an equally sensitive measure of a coarse, global environmental representation. During Planning, activation in the hippocampal head was also associated with the Direction factor. The region in the hippocampal head that was related to the Direction factor was also related to Direction to remote landmarks. This indicates that the hippocampal head is important when direction is evaluated using a global environmental frame. In summary, we find the hippocampal head to be especially important for supporting a coarse, global representation of the environment. Although previous fMRI studies have not investigated this issue explicitly, several studies are consistent with the current findings. For example, activation in the hippocampal head has been observed to increase with increasing distance between real world landmarks (Morgan et al., 2011) and in persons who use relative distance to more than two landmarks to reach four target landmarks (Iaria, Petrides, Dagher, Pike, & Bohbot, 2003). These findings indicate that the hippocampal head is sensitive to large-scale features of the environment and supports the ability to form a coarse, global environmental representation during learning.

Our structural image analysis also supported a role for the hippocampal head in representation of a coarse, global representation. Specifically, we found that use of a coarse, global representation of the environment during Self-localization and Planning correlated with a larger right hippocampal head and smaller right hippocampal tail. Hence, involvement of the hippocampal head in a coarse, global environmental representation was not only reflected in the level of activity, but also in its structure. The fact that the hippocampal tail was smaller in participants who reported greater use of a coarse, global environmental representation, strengthens the claim that the coarse, global environmental representation in the hippocampal head is independent of the fine-grained, local representations in the hippocampal tail. Supporting this, blind individuals were observed to have a larger hippocampal head and an increased ability to recognize a global overview representation of learned environments (Leporé et al., 2009; Fortin et al., 2008). We found no relationship between the spatial tests and local hippocampal shape. This is similar to a study of London taxi drivers, where no relationship was found between hippocampal shape and the ability to judge landmark proximity (Woollett & Maguire, 2011). The reason for the lack of relationship in this and in our study is not clear. However, one could speculate that strategy questionnaire factors are more sensitive to personality traits than test scores. Personality traits are known to influence the shape of brain structures (Gardini, Cloninger, & Venneri, 2009).

Our findings confirmed the prediction that fine-grained, local representations are processed in the hippocampal tail. During Planning, activation in the hippocampal tail correlated with the Local and Global direction test scores, but not with the Multiple distance scores. The main difference between the direction tests and the distance test is that the former required a fine-grained mental representation (of the exact position and orientation) of the landmarks' position relative to the adjacent walls, whereas the latter required a coarse, global representation of the environment with the different parts of the environment merged and oriented relative to one another. Thus, our finding indicates that the hippocampal tail is important for fine-grained, local environmental representations. The behavioral analysis showed that activation in the hippocampal tail correlated with the questionnaire factors Sequence and turns and Direction during Planning, but not with the factors Direction to remote landmarks or Map. The first two factors reflect the use of fine-grained representations of the local environment. The results are consistent with the observation that the hippocampal tail is important for the exact representation of an individual object's location (Baumann, Chan, & Mattingley, 2010; Rodriguez, 2010; Doeller, King, & Burgess, 2008). Other studies have linked activation in the hippocampal tail to choosing the optimal environmental path (Hartley, Maguire, Spiers, & Burgess, 2003) and to the amount of details recalled or imagined from past and future events (Addis & Schacter, 2008). Furthermore, there is evidence that the hippocampal tail represents local parts of the environment (Xu et al., 2010; Janzen & Weststeijn, 2007; Peigneux et al., 2004), such as decision point objects (Janzen & Weststeijn, 2007). Our study extends these findings by providing the first direct evidence that the human hippocampal tail represents the local environment most precisely, as previously shown in rats (Kjelstrup et al., 2008; Jung et al., 1994).

The volume in the hippocampal tail did not correlate positively with use of fine-grained, local environmental representations. This could be related to a smaller variance being observed for the fine-grained, local aggregate score (range = 15–49) than the coarse, global aggregate score (range = 17–34).

The inferior parietal lobe appeared to have a similar posterior–anterior segregation based on the self-reported use of a fine-grained, local route representation and/or a global route representation. During Planning, increased activation of the left inferior parietal lobe was observed in one anterior and one posterior cluster. The posterior activation was associated with relying on number of turns between the start position and target landmark, and the anterior cluster with the use of complete routes. In the right hemisphere, activation in the posterior inferior parietal lobe was related to using sequences of landmarks. These findings suggest that the inferior parietal cortex supports route-centered representations by processing fine-grained, local route representations posteriorly and a global route representation anteriorly. Activation in the inferior parietal cortex has previously been associated with spontaneous route planning (Spiers & Maguire, 2006) and correlated with the ability to tell the spatial relationship between landmarks in a route centered reference frame (Wolbers, Weiller, & Büchel, 2004). Our results add to these findings by suggesting a functional specialization within the inferior parietal cortex that reflects the completeness of environmental route representations. Past research has only provided indirect support for such a distinction. For example, the posterior inferior parietal cortex has been associated with object-specific responses (Konen & Kastner, 2008a) and saccadic eye movements (Konen & Kastner, 2008b). Sacccadic eye movements are necessary to remember local details from images (Loftus, 1972), but not overviews of spatial scenes and objects (Potter, 1976).

In conclusion, our study shows that a coarse, global representation of the environment is established in the hippocampal head, whereas the hippocampal tail supports fine-grained, local environmental representations. The anterior part of the inferior parietal cortex supports a coarse, global route-representation, whereas fine-grained, local route representations are found posteriorly. Further studies focusing on a functional segregation along the anterior–posterior axis both in the hippocampus and parietal cortex as well as in other regions of the brain will be important, not only to understand the nature of neural processing in these regions but also to elucidate fundamental organizational principles of the human brain.

Acknowledgments

This work was supported by a grant from the Norwegian Research Council. We also thank the staff at the Department of Medical Imaging at St. Olavs Hospital in Trondheim for assistance with imaging protocols and data acquisition.

Reprint requests should be sent to Hallvard Røe Evensmoen, Department of Neuroscience, Faculty of Medicine, Norwegian University of Science and Technology, 7489 Trondheim, Norway, or via e-mail: Hallvard.r.evensmoen@ntnu.no.

REFERENCES

REFERENCES
Addis
,
D. R.
, &
Schacter
,
D. L.
(
2008
).
Constructive episodic simulation: Temporal distance and detail of past and future events modulate hippocampal engagement.
Hippocampus
,
18
,
227
237
.
Barra
,
J.
,
Laou
,
L.
,
Poline
,
J.-B.
,
Lebihan
,
D.
, &
Berthoz
,
A.
(
2012
).
Does an oblique/slanted perspective during virtual navigation engage both egocentric and allocentric brain strategies?
PLoS ONE
,
7
,
e49537
.
Baumann
,
O.
,
Chan
,
E.
, &
Mattingley
,
J. B.
(
2010
).
Dissociable neural circuits for encoding and retrieval of object locations during active navigation in humans.
Neuroimage
,
49
,
2816
2825
.
Beckmann
,
C. F.
,
Jenkinson
,
M.
, &
Smith
,
S. M.
(
2003
).
General multilevel linear modeling for group analysis in fMRI.
Neuroimage
,
20
,
1052
1063
.
Blanch
,
R. J.
,
Brennan
,
D.
,
Condon
,
B.
,
Santosh
,
C.
, &
Hadley
,
D.
(
2004
).
Are there gender-specific neural substrates of route learning from different perspectives?
Cerebral Cortex
,
14
,
1207
1213
.
Boynton
,
G. M.
,
Engel
,
S. A.
,
Glover
,
G. H.
, &
Heeger
,
D. J.
(
1996
).
Linear systems analysis of functional magnetic resonance imaging in human V1.
Journal of Neuroscience
,
16
,
4207
4221
.
Colle
,
H. A.
, &
Reid
,
G. B.
(
2000
).
The room effect: Exploring paths and rooms in a desktop virtual environment with objects grouped categorically and spatially.
Ecological Psychology
,
12
,
207
229
.
De Winter
,
J. C. F.
,
Dodou
,
D.
, &
Wieringa
,
P. A.
(
2009
).
Exploratory factor analysis with small sample sizes.
Multivariate Behavioral Research
,
44
,
147
181
.
DeFelipe
,
J.
,
Fernández-Gil
,
M. Á.
,
Kastanauskaite
,
A.
,
Palacios Bote
,
R.
,
Gañán Presmanes
,
Y.
, &
Ruiz
,
M. T.
(
2007
).
Macroanatomy and microanatomy of the temporal lobe.
Seminars in Ultrasound, CT, and MRI
,
28
,
404
415
.
Doeller
,
C. F.
,
King
,
J. A.
, &
Burgess
,
N.
(
2008
).
Parallel striatal and hippocampal systems for landmarks and boundaries in spatial memory.
Proceedings of the National Academy of Sciences
,
105
,
5915
5920
.
Duvernoy
,
H. M.
(
2005
).
The human hippocampus: Functional anatomy, vascularization, and serial sections with MRI.
Berlin
:
Springer-Verlag
.
Fortin
,
M.
,
Voss
,
P.
,
Lord
,
C.
,
Lassonde
,
M.
,
Pruessner
,
J.
,
Saint-Amour
,
D.
,
et al
(
2008
).
Wayfinding in the blind: Larger hippocampal volume and supranormal spatial navigation.
Brain
,
131
,
2995
.
Gardini
,
S.
,
Cloninger
,
C. R.
, &
Venneri
,
A.
(
2009
).
Individual differences in personality traits reflect structural variance in specific brain regions.
Brain Research Bulletin
,
79
,
265
270
.
Grön
,
G.
,
Wunderlich
,
A.
,
Spitzer
,
M.
,
Tomczak
,
R.
, &
Riepe
,
M.
(
2000
).
Brain activation during human navigation: Gender-different neural networks as substrate of performance.
Nature Neuroscience
,
3
,
404
408
.
Hartley
,
T.
,
Maguire
,
E. A.
,
Spiers
,
H. J.
, &
Burgess
,
N.
(
2003
).
The well-worn route and the path less traveled: Distinct neural bases of route following and wayfinding in humans.
Neuron
,
37
,
877
888
.
Iaria
,
G.
,
Petrides
,
M.
,
Dagher
,
A.
,
Pike
,
B.
, &
Bohbot
,
V.
(
2003
).
Cognitive strategies dependent on the hippocampus and caudate nucleus in human navigation: Variability and change with practice.
Journal of Neuroscience
,
23
,
5945
5952
.
Insausti
,
R.
,
Juottonen
,
K.
,
Soininen
,
H.
,
Insausti
,
A.
,
Partanen
,
K.
,
Vainio
,
P.
,
et al
(
1998
).
MR volumetric analysis of the human entorhinal, perirhinal, and temporopolar cortices.
American Journal of Neuroradiology
,
19
,
659
671
.
Janzen
,
G.
, &
Weststeijn
,
C. G.
(
2007
).
Neural representation of object location and route direction: An event-related fMRI study.
Brain Research
,
1165
,
116
125
.
Jung
,
M.
,
Wiener
,
S.
, &
McNaughton
,
B.
(
1994
).
Comparison of spatial firing characteristics of units in dorsal and ventral hippocampus of the rat.
Journal of Neuroscience
,
14
,
7347
7356
.
Kaza
,
E.
,
Klose
,
U.
, &
Lotze
,
M.
(
2011
).
Comparison of a 32-channel with a 12-channel head coil: Are there relevant improvements for functional imaging?
Journal of Magnetic Resonance Imaging
,
34
,
173
183
.
Kjelstrup
,
K. B.
,
Solstad
,
T.
,
Brun
,
V. H.
,
Hafting
,
T.
,
Leutgeb
,
S.
,
Witter
,
M. P.
,
et al
(
2008
).
Finite scale of spatial representation in the hippocampus.
Science
,
321
,
140
143
.
Konen
,
C. S.
, &
Kastner
,
S.
(
2008a
).
Two hierarchically organized neural systems for object information in human visual cortex.
Nature Neuroscience
,
11
,
224
231
.
Konen
,
C. S.
, &
Kastner
,
S.
(
2008b
).
Representation of eye movements and stimulus motion in topographically organized areas of human posterior parietal cortex.
The Journal of Neuroscience
,
28
,
8361
8375
.
Leporé
,
N.
,
Shi
,
Y.
,
Lepore
,
F.
,
Fortin
,
M.
,
Voss
,
P.
,
Chou
,
Y.-Y.
,
et al
(
2009
).
Pattern of hippocampal shape and volume differences in blind subjects.
Neuroimage
,
46
,
949
957
.
Loftus
,
G. R.
(
1972
).
Eye fixations and recognition memory for pictures.
Cognitive Psychology
,
3
,
525
551
.
Lynch
,
K.
(
1960
).
The image of the city.
Cambridge, MA
:
MIT Press
.
Morgan
,
L. K.
,
MacEvoy
,
S. P.
,
Aguirre
,
G. K.
, &
Epstein
,
R. A.
(
2011
).
Distances between real-world locations are represented in the human hippocampus.
Journal of Neuroscience
,
31
,
1238
1245
.
Nadel
,
L.
,
Hoscheidt
,
S.
, &
Ryan
,
L. R.
(
2012
).
Spatial cognition and the hippocampus: The anterior-posterior axis.
Journal of Cognitive Neuroscience
,
25
,
22
28
.
Newcombe
,
N.
, &
Liben
,
L. S.
(
1982
).
Barrier effects in the cognitive maps of children and adults.
Journal of Experimental Child Psychology
,
34
,
46
58
.
Nitz
,
D. A.
(
2006
).
Tracking route progression in the posterior parietal cortex.
Neuron
,
49
,
747
756
.
Nitz
,
D.
(
2009
).
Parietal cortex, navigation, and the construction of arbitrary reference frames for spatial information.
Neurobiology of Learning and Memory
,
91
,
179
185
.
Ohnishi
,
T.
,
Matsuda
,
H.
,
Hirakata
,
M.
, &
Ugawa
,
Y.
(
2006
).
Navigation ability dependent neural activation in the human brain: An fMRI study.
Neuroscience Research
,
55
,
361
369
.
Peigneux
,
P.
,
Laureys
,
S.
,
Fuchs
,
S.
,
Collette
,
F.
,
Perrin
,
F.
,
Reggers
,
J.
,
et al
(
2004
).
Are spatial memories strengthened in the human hippocampus during slow wave sleep?
Neuron
,
44
,
535
545
.
Potter
,
M. C.
(
1976
).
Short-term conceptual memory for pictures.
Journal of Experimental Psychology: Human Learning and Memory
,
2
,
509
522
.
Rodriguez
,
P. F.
(
2010
).
Human navigation that requires calculating heading vectors recruits parietal cortex in a virtual and visually sparse water maze task in fMRI.
Behavioral Neuroscience
,
124
,
532
540
.
Serences
,
J. T.
, &
Yantis
,
S.
(
2006
).
Selective visual attention and perceptual coherence.
Trends in Cognitive Sciences
,
10
,
38
45
.
Siegel
,
A. W.
, &
White
,
S. H.
(
1975
).
The development of spatial representations of large-scale environments.
In W. R. Hayne (Ed.)
,
Advances in child development and behavior
(pp.
9
55
).
New York
:
JAI
.
Silver
,
M. A.
, &
Kastner
,
S.
(
2009
).
Topographic maps in human frontal and parietal cortex.
Trends in Cognitive Sciences
,
13
,
488
495
.
Smith
,
S. M.
,
Jenkinson
,
M.
,
Woolrich
,
M. W.
,
Beckmann
,
C. F.
,
Behrens
,
T. E. J.
,
Johansen-Berg
,
H.
,
et al
(
2004
).
Advances in functional and structural MR image analysis and implementation as FSL.
Neuroimage
,
23(Suppl. 1)
,
S208
S219
.
Spiers
,
H. J.
, &
Maguire
,
E. A.
(
2006
).
Thoughts, behaviour, and brain dynamics during navigation in the real world.
Neuroimage
,
31
,
1826
1840
.
Wang
,
R. F.
, &
Brockmole
,
J. R.
(
2003
).
Human navigation in nested environments.
Journal of Experimental Psychology: Learning, Memory, and Cognition
,
29
,
398
404
.
Wolbers
,
T.
,
Weiller
,
C.
, &
Büchel
,
C.
(
2004
).
Neural foundations of emerging route knowledge in complex spatial environments.
Cognitive Brain Research
,
21
,
401
411
.
Woollett
,
K.
, &
Maguire
,
E. A.
(
2011
).
Acquiring “the knowledge” of London's layout drives structural brain changes.
Current Biology
,
21
,
2109
2114
.
Xu
,
J.
,
Evensmoen
,
H. R.
,
Lehn
,
H.
,
Pintzka
,
C. W. S.
, &
Håberg
,
A. K.
(
2010
).
Persistent posterior and transient anterior medial temporal lobe activity during navigation.
Neuroimage
,
52
,
1654
1666
.