Evidence for the Use of an Internal Sense of Direction in Homing

Matthijs A. A. van der Meer, Zoe Richmond,Rodrigo M. Braga, and Emma R. Wood

University of Edinburgh

Paul A. DudchenkoUniversity of Edinburgh and Stirling University

Path integration, the ability to maintain a representation of location and direction on the basis of internalcues, is thought to be important for navigation and the learning of spatial relationships. Representationsof location and direction in the brain, such as head direction cells, grid cells, and place cells in the limbicsystem, are thought to underlie navigation by path integration. While this idea is generally consistent withlesion studies, the relationship between such neural activity and behavior has not been studied on a taskwhere animals demonstrably use a path integration strategy. Here we report the development of such atask in rats: by slowly rotating rats before their return to a trial-unique home base, we could show subjectsrelied on internal cues only to navigate. To illustrate how this task can be combined with recording, weshow examples of simultaneously recorded head direction cells in which neural activity is closely relatedto rats’ homing direction. These results support the notion that rats can navigate by path integration, thatthis ability depends on head direction cells, and suggest a convenient behavioral paradigm for investi-gating the neural basis of navigation.

Keywords: path integration, head direction, navigation, rat

Supplemental materials: http://dx.doi.org/10.1037/a0018446.supp

Navigation is an important survival skill for many mammals,such as for rats returning to their nest after a foraging trip, or forhumans attempting to find their car in the parking lot. A majorsubdivision between navigation strategies is that between thoserelying on external cues, such as landmarks in the environment,and those relying on internal cues, such as vestibular or proprio-ceptive feedback (O’Keefe and Nadel, 1978; Schone, 1984; Gal-listel, 1990; Redish, 1999). The latter strategy is referred to as“dead reckoning” or path integration, and is thought to be impor-tant, not only when external cues are unavailable or unreliable, butalso for learning spatial relationships, such as the construction of“cognitive maps” (McNaughton, Battaglia, Jensen, Moser, &Moser, 2006). Demonstrating path integration experimentally canbe difficult because of the variety of potential external-cue-based

alternative strategies that need to be ruled out (Jeffery & Etienne,2004). In rodents, a number of tasks have been developed thatencourage a path integration strategy by attempting to minimizethe availability or informativeness of external cues (e.g., Etienne,Maurer, Saucy, & Teroni, 1986; Whishaw & Tomie, 1997; Pearce,Roberts, & Good, 1998); the fact that subjects typically still dowell in such tasks (although usually not as well compared to whenlandmarks are available) supports the notion that they are unlikelyto be relying on external cues. A complementary approach hasbeen to analyze subtle aspects of navigation behavior, such asstructural biases in homing (suggesting path integration errors,Seguinot, Maurer, & Etienne, 1993) or detailed speed profiles(suggesting rats know the distance to a home box even on theirfirst outbound journey, Wallace & Whishaw, 2003). Taken to-gether, these results support the idea that rodents can navigate onthe basis of internal cues alone.

However, the neural basis of path integration remains the sub-ject of debate. Theoretically, a combination of direction and dis-tance information is sufficient to plan a direct path between anytwo points, leading to the proposal that “head direction cells”(Taube, 2007), and “place cells” or “grid cells” (Moser, Kropff, &Moser, 2008) in the limbic system might support path integration(McNaughton, Chen, & Markus, 1991). While this notion hasgenerally been supported by lesion studies (e.g., Wallace &Whishaw, 2003; Wilton, Baird, Muir, Honey, & Aggleton, 2001;Vann, 2005; Frohardt, Bassett, & Taube, 2006; but see Alyan &McNaughton, 1999; Shrager, Kirwan, & Squire, 2008), attempts torelate neural activity to spatial behavior have yielded contradictoryresults (Golob, Stackman, Wong, & Taube, 2001; Muir & Taube,2004; Jeffery, Gilbert, Burton, & Strudwick, 2003; Lenck-Santini,Muller, Save, & Poucet, 2002). A limitation of these studies is thatthey do not provide evidence that rats are in fact path integrating,raising the possibility that they are solving tasks by strategies that

Matthijs A. A. van der Meer, Centre for Cognitive and Neural Systems,Neuroinformatics Doctoral Training Centre, University of Edinburgh, Ed-inburgh, U.K.; Zoe Richmond, Rodrigo M. Braga, and Emma R. Wood,Centre for Cognitive and Neural Systems, University of Edinburgh, Edin-burgh, U.K.; and Paul A. Dudchenko, Centre for Cognitive and NeuralSystems, University of Edinburgh, Edinburgh, U.K., and Department ofPsychology, Stirling University, Stirling, U.K.

Supported by an EPSRC/MRC Neuroinformatics Doctoral TrainingCentre studentship to MvdM. We thank Hugh Cameron, Douglas Howie,and Robert MacGregor for apparatus design and construction, Kate Shiresfor help with running animals, Rosamund Langston and James Ainge forassistance with surgery, Jane Tulloch for histology, and Neil Burgess,Mayank Dutia, A. David Redish, and Mark van Rossum for discussion.This work has been previously published in abstract form (Society forNeuroscience annual meeting, 2007).

Correspondence concerning this article should be addressed to MatthijsA. A. van der Meer, Department of Neuroscience, University of Minnesota,6-145 Jackson Hall, 321 Church St. SE, Minneapolis, MN 55455. E-mail:mvdm@umn.edu

Behavioral Neuroscience © 2010 American Psychological Association2010, Vol. 124, No. 1, 164–169 0735-7044/10/$12.00 DOI: 10.1037/a0018446

164

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

do not require representations of direction or place (Dudchenko,Muir, Frohardt, & Taube, 2005). There has been no demonstrationof head direction or place cell activity on a task which explicitlytests for a path integration strategy.

Arguably the strongest demonstration of path integration in-volves active manipulation, such as rotation below the vestibularthreshold, of the subject (Jeffery & Etienne, 2004). By pitting thesubject’s (manipulated) internal cues against (stable) external cues,path integration can be dissociated from navigation based onexternal cues. In a classic experiment, Mittelstaedt and Mittel-staedt (1980) found that female desert mice retrieving theiryoung would return to a location that matched the preciseamount they had been rotated on their outbound journey, show-ing that they were relying on internal cues. This result has notbeen shown in rats, the model system of choice for studying theneural representations of path integration. This would be animportant development as, by virtue of the rotation manipula-tion, which demonstrates the use of internal cues, this task canprovide a basis for investigating the neural representationsunderlying path integration. Thus, we sought to first, develop aversion of Mittelstaedt and Mittelstaedt (1980)’s experiment inrats, using appetitive rewards, and second, demonstrate that insuch a setting—where rats are demonstrably using a path inte-gration strategy—activity in head direction cells can be verytightly linked to behavior.

Method

We developed a homing task designed to test rats’ ability toreturn to a home location using internal (idiothetic) cues only. Rats(n � 12 for behavior, n � 3 for unit recording) were trained toretrieve food pellets from the center of an illuminated circular openfield (185 cm in diameter, Figure 1A), returning to a trial-uniquehome location (a removable “home box”) to eat. Rats were gentlydisoriented before all training and probe trial sessions. On probetrials, rats could be confined to the center (by raising a cylindricaltube 25 cm in diameter) for 60 seconds, while the table wascleaned, the home box replaced, and three additional home boxesinserted at 90° intervals relative to the original home location. Onrotation probe trials, the center was additionally rotated 90° clock-wise or counterclockwise during confinement. After tube lower-ing, the rats’ first choice of return location was recorded; thisprocedure allowed us to distinguish whether rats were using stableexternal cues or an internal sense to return home. If rats were usingexternal cues, they should not be affected by rotation, since thecues have not moved relative to the home location. In contrast, ifthey were using an internal sense of direction (and the rotation issuch that it cannot be compensated for) they should be affected; inthe case where they are fully unaware of being rotated, their choiceof return should correspond to the rotation (Figure 1B). Full detailsof the training protocol, data collection and unit recording proce-

no rotation trial

A

rotation trial

return choice

?

B

32

no rotation72 trials, |r| = 0.44 (6.9°)

25

90° rotation 90 trials, |r| = 0.31 (78.2°) no rot rot

0

0.1

0.2

0.3

0.4

0.5

first

cho

ice

corr

ect

* *

C D E

rat enters centerto retrieve food

A

Figure 1. Rats can successfully return to a trial-unique starting location after confinement, and rotations shifttheir return choice. A: The apparatus was a large circular table, illuminated from above, with removable startboxes evenly spaced around the perimeter, and a rotatable platform in the center. Top photo shows an overheadview, bottom photo the view from a start box. Rats were trained to fetch food pellets from the center of the tableand return to a trial-unique starting location. B: Task design schematic. On probe trials, before making theirreturn choice, rats were confined to the center of the table, which was either rotated (rotation trials: 90° over 60 s)or not rotated. C: Circular histograms of the first location rats chose to return to on no-rotation probe trials. Alltrials are aligned to the same starting location, indicated by the black asterisk at 0°. Rats home significantly tothe home location of 0°. D: Homing performance on rotation probe trials, using the same layout as the left panel.Trials are aligned to the same starting location (black dot), and clockwise rotation trials plotted in reverse to becomparable to counterclockwise trials. In contrast to C, rats homed to the rotated location. E: The proportion oftrials on which rats chose the correct box (0° on nonrotation, rotated box on rotation trials) was not affected bythe rotation.

165BRIEF COMMUNICATIONS

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

dures, and analysis can be found in the Supplementary OnlineMaterials.

Results

Rats (n � 12) ran a homing task where they could be confinedand rotated before returning to their starting location. On nonro-tation probe trials, rats’ return choices were biased toward a singlemean homing direction (Figure 1C; Rayleigh test for uniformity,r � .44; p � .001) consistent with a true homing angle of 0°(mean � 99% confidence interval: 6.9° � 25.3°), indicating thatrats were able to return to a trial-unique starting location afterconfinement. For analysis of rotation trials (Figure 1D) data fromclockwise trials were reversed to be comparable with counter-clockwise trials. After rotation, choices were significantly biasedtoward a direction consistent with population mean of 90° (homingdirection 78.2° � 35.3°, Rayleigh test: r � .31; p � .001). Thehypothesis of a common mean direction between the nonrotationand rotation probe trials could be rejected (Watson’s Yr � 17.6,p � .001). Thus, rats were affected by the rotation, tending toreturn to the location corresponding to the rotation, without anoverall decrease in homing performance (percentage of correctfirst choices, Figure 1E), demonstrating that they are not usingstable external cues to guide their behavior and rely on internalcues instead.

Neural activity from head direction (HD) cells was recorded asrats (n � 3) performed the homing task described above. A total of11 sessions which yielded at least one cell which met the inclusioncriteria of significant directional tuning (Supplementary OnlineMaterials) were recorded. Cells were recorded from the anteriorthalamic nucleus (1 cell), the postsubiculum (5 cells) and the CA1region of the hippocampus1 (2 cells). A video recording of rat 0304performing a rotation trial can be seen in the SupplementaryOnline Materials. To examine the relationship between HD cellactivity and behavior, we compared HD cells’ preferred firingdirection (PFD) shift (the angular difference between PFDs pre-and post-confinement) to rats’ choice of return location. Resultsfrom an example individual trial, with two simultaneously re-corded HD cells, are shown in Figure 2A. Before confinement, therat’s home box was located in the northwest (red dot). Afterconfinement (with no rotation taking place), the rat took the thickblue path from the center, corresponding to a choice of �19°(indicated by the blue dot; positive numbers are counterclockwise),ending up at its starting location. An example rotation trial isshown in Figure 2B. In both these cases, there was close agreementbetween the change in preferred firing direction of two simulta-neously recorded HD cells and the rat’s choice of return path. Wealso observed trials in which a HD cell PFD shift occurred withoutrotation of the center tube; this and further examples can be foundin the Supplementary Online Materials.

Not all sessions showed such close alignment between HD cellPFD and behavior; in two sessions, we observed an apparentmismatch. However, in one such case (Figure S2) this may havebeen due to the inherent difficulty in estimating HD cell PFDs,which may change over time and are subject to sampling bias.Regardless of whether this ambiguous data point was included,corrected, or excluded, there was a significant correlation betweenHD cell PFD and choice of return location (Figure 2C; T-linearcircular correlation (Fisher, 1993), r � .46; p � .001 when

excluded). This panel shows the rat’s choice of return plottedagainst changes in cell PFDs, for all trials from three rats. Theambiguous session in Figure S2 is indicated in its two possiblepositions, with the other mismatch sessions also labeled. Each datapoint has three further dimensions apart from its position on theplot. Color indicates which rat the cell was recorded from; eachsession has a unique shape, and their size is proportional to tuningvector length. Thus, the three small five-pointed blue stars on theleft of the plot are cells recorded simultaneously from rat 0304, andwhich are not as well tuned for direction as most other cells. Thedotted line indicates x � y, where behavioral choice and cell PFDchange are equal; most data points lie on or close to the dashedline, indicating agreement between behavioral choice and PFDchange. A different measure quantifying the relationship betweenPFD shift and choice can be obtained by comparing the distribu-tion of differences between PFD shift and behavior to the levelexpected if the two were independent, that is, 90° different onaverage. The observed average difference (43.3° � 9.4°, ambigu-ous data point removed) is significantly below this chance level(Student’s one-sample t test: t(16) � �4.9; p � .001). Including theuncorrected ambiguous data does not substantially alter this result(t(18) � �3.17; p � .005). Thus, even in this limited data set, thereis evidence of a link between the HD signal and the rats’ behavior.

Discussion

We found that on a homing task, rats’ choice of return pathcould be manipulated by slow rotations before they begin theirreturn journey, implying that they are not using stable externalcues to navigate, and must be relying on internal (idiothetic) cues,or path integrating, instead. This approach to demonstrating pathintegration is different from previous work in rats, which hastypically relied on minimizing the availability or the informative-ness of cues in the environment (Etienne et al., 1986; Whishaw &Tomie, 1997; Pearce et al., 1998). While such studies have gen-erally supported the notion that rats can use path integrationstrategies to navigate, it can be difficult to exclude the possibilitythat some uncontrolled cue is being used for navigation, as rats areknown to be able to use many different types of cues to navigate(Maaswinkel & Whishaw, 1999; Wallace, Gorny, & Whishaw,2002). The present study avoids this issue by showing that rota-tions of the animal result in a corresponding change in homingbehavior, an approach pioneered by Mittelstaedt and Mittelstaedt(1980) in female desert mice retrieving their young. The interpre-tation that on this task, rats are not using external cues in theenvironment is further illustrated by anecdotal observations of ratsjust missing the (correct) home location by a small distance, andthen turning the wrong way, implying they could not smell orsense the presence of a home box in that location. Accordingly, interms of choices, rats did not prefer the � 90° locations (i.e., thosecontaining boxes) over the average of the � 45° and � 135°locations (without boxes; �2 � 1.96; df � 1; p � .16) on probetrials. Similarly, over all trials, there was no evidence of a bias

1 Although relatively rare, several reports of HD recordings have beennoted in this area, for instance by Ranck Jr. (2005) and Leutgeb,Ragozzino, and Mizumori (2000). It is possible these recordings were infact from the axons of HD cells in entorhinal cortex.

166 BRIEF COMMUNICATIONS

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

toward particular locations in the stable room frame (Rayleigh test:r � .05; p � .63). Thus, by using a rotation manipulation, thepresent results provide novel evidence that rats are able to navigateby path integration.

Although the rotation manipulation reliably shifted the rats’return choices in a manner consistent with a path integrationstrategy, rats returned to the “correct” home box about 35% of thetime. While comfortably above chance, this number indicatessignificant limitations on how well rats were able to perform thetask. There are several possible contributions that could accountfor this: first, the rats were confined for 60 seconds in a tube thatwas raised and lowered in close proximity, placing working mem-ory demands alongside a distracting and mildly stressful event.Thus it is possible that on some trials, rats were unable to remem-ber the goal location over this interval, consistent with previousexplanations for behavioral variability in “shortcut” tasks (Tolman,Ritchie, & Kalish, 1946; Muir & Taube, 2004). Another possibilityis that homing performance was limited by constraints on the pathintegration process itself, especially because rats frequently movedaround frantically during confinement; this idea is supported byobservations of rats homing to a location consistent with a shift inhead direction cell preferred direction (discussed below). Finally,it is possible that rats were not always motivated to return to a

home location which was never explicitly rewarded. However,when we ran three subjects on a version of the task where rats werestarted from the same location, and several cues were provided onthe apparatus and surrounding curtain, homing performance wasclose to 70% correct (Figure S4). This suggests that motivationalfactors were not a strong constraint on homing performance.

The current task provides a convenient paradigm for studyingthe neural basis of path integration. To illustrate this, we show thaton this task, the PFD of HD cells could change in very closeagreement with the rats’ chosen homing direction. Such agreementwas observed on nonrotation and rotation probe trials, as well asduring error trials where the rats returned to neither their originalnor the rotated home location. This suggests that such matchesbetween HD cell PFD and behavior were not due to alignmentsattributable to external cues, where both the HD cells and behavioralign to the same stable cue. Moreover, in cases where multipleHD cells were recorded simultaneously, the cells’ PFD shifts wereconsistent, further reducing the likelihood that such alignmentsoccurred by chance. Thus, these data demonstrate that under pathintegration conditions, the “neural compass” thought to be pro-vided by HD cells, can be closely aligned with rats’ homingdirection. The fact that we also observed mismatches between

session 4322_r - s (no rotation)choice: -19o

PRE (inward) / POST (return)

cell 1PFD shift: -32o

80Hz

cell 2PFD shift: -3o

30HzA

session 4322_s - t (90o rotation)choice: 78o

PRE (inward) / POST (return)

cell 1PFD shift: 83o

105Hz

cell 2PFD shift: 86o

30HzB

C

-180 -90 0 90 180-180

-90

0

90

180

Return choice (deg)

Cel

l PF

D r

otat

ion

(deg

)

rat 4322

rat 6225

rat 0304

1

1

1ambiguous session, see main text

r = 0.46, p = 0.001 (1excluded)

r = 0.50, p < 0.001 (1corrected)

r = 0.30, p = 0.002 (1uncorrected)

Figure 2. Head direction (HD) cell activity correlated with behavioral choice. A: Example nonrotation session,where the rat started from the northwest location (red dot, red lines indicate preconfinement position), and afterconfinement in the center of the maze, returned to �19° (blue dot, thick blue line indicates initial return).Postconfinement position sampling is indicated in thin blue. Tuning curves of head direction cells recordedduring this trial, with the preconfinement tuning curve in red and the postconfinement tuning curve in blue, areshown on the right. For the two cells respectively, preferred firing directions (PFDs) shifted �32° and �3°, inrelatively close agreement with behavioral choice. B: Example counterclockwise rotation trial, with panel layoutas in A. Again, the PFD shifts of 83° and 86° were in agreement with behavioral choice (78°). C: Scatterplot ofreturn choice against shift in preferred direction of recorded HD cells, for all recording sessions. Each pointcorresponds to a recording of a single HD cell during a single session. Each data point has three furtherdimensions apart from its position on the plot: color indicates which rat the cell was recorded from, differentshapes correspond to different recording sessions, and size is proportional to tuning vector length (larger shapeindicates better tuning). The dashed line indicates x � y; PFD shifts are significantly correlated with behavioralchoice.

167BRIEF COMMUNICATIONS

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

behavior and HD cell PFDs could result from inaccurate memoryfor the home location on a particular trial, rather than from inac-curate navigation to the correctly remembered home.

The observation that slow rotations can change the preferredfiring direction of HD cells (thought to result from vestibularadaptation, van der Meer, Knierim, Yoganarasimha,Wood, & vanRossum, 2007) is consistent with previous results (in place cells,Jeffery, Donnett, Burgess, & O’Keefe, 1997; Hargreaves, Yoga-narasimha, & Knierim, 2007), although these studies did not relatesuch changes to subsequent spatial behavior. Previous work inves-tigating the relationship between the HD signal and behavior(Mizumori & Williams, 1993; Dudchenko & Taube, 1997; Golobet al., 2001; Frohardt, Marcroft, & Taube, 2002; Muir & Taube,2004) relied on manipulation of the HD signal by changing cues inthe environment. Results from these studies have been conflicting,with some reporting correlations between HD cell activity andbehavior, and others finding dissociations (Muir & Taube, 2002;Dudchenko et al., 2005). However, in these studies, there is noevidence that rats require a directional signal to solve the task, suchthat the absence of a correlation may simply reflect the use of adifferent (non-HD dependent) strategy. The present study avoidsthis interpretation by demonstrating rats are relying on internalsignals to return to their home location.

While we recorded HD cells from several different areas, itwas not our aim to compare their differences. Although thereare differences in the properties of head direction cells recordedfrom different areas (e.g., in the amount by which they antici-pate true HD, the width of their tuning curve, or their relation-ship to angular and linear velocity; Taube & Bassett, 2003) theonly property we rely on in our analysis is that the cells have astable directional tuning curve. This has been consistentlyfound between different HD-cell containing areas: while weknow of no simultaneous recordings of HD cells in multipleareas, simultaneous HD and place cell recordings suggest thatthe directional properties of these cells are consistent betweenthe anterodorsal thalamic nucleus (ADN) and the hippocampus(Knierim, Kudrimoti, & McNaughton, 1998). Since the post-subicular HD signal depends on an intact ADN (Goodridge &Taube, 1997), and hippocampal firing stability depends on thepostsubiculum (Calton et al., 2003) it is likely that these threeareas form an integrated system that shares a consistent repre-sentation of head direction. The number of recordings in thisstudy was unfortunately limited because of the departure of theauthor responsible for this part of the study. It is possible thata more extensive comparison of the properties of HD cells indistinct brain areas would reveal differences between recordingsites; however, our observation that across areas, there tendedto be a significant relationship between HD cell PFD changesand behavior supports the notion that on a path integration task,rats can rely on the directional signal provided by the HDsystem.

In summary, we report that on a homing task where rats de-monstrably rely on a path integration strategy, there are clearinstances where simultaneously recorded HD cells closely align tobehavioral choice. This suggests the HD system is relevant to pathintegration. This paradigm lends itself to extension to other spatialrepresentations.

References

Alyan, S., & McNaughton, B. L. (1999). Hippocampectomized rats arecapable of homing by path integration. Behavioral Neuroscience, 113,19–31.

Calton, J. P., Stackman, R. W., Goodridge, J. P., Archey, W., Dudchenko,P. A., & Taube, J. S. (2003). Hippocampal place cell instability afterlesions of the head direction cell network. Journal of Neuroscience, 23,9719–9731.

Dudchenko, P. A., Muir, G., Frohardt, R., & Taube, J. S. (2005). What doesthe head direction system actually do? In S. Wiener, and J. Taube (Eds.),Head direction cells and the neural mechanisms of spatial orientation,pp. 221–246. Cambridge, MA: MIT Press.

Dudchenko, P. A., & Taube, J. S. (1997). Correlation between headdirection cell activity and spatial behavior on a radial arm maze. Behav-ioral Neuroscience, 111, 3–19.

Etienne, A. S., Maurer, R., Saucy, F., & Teroni, E. (1986). Short-distancehoming in the golden hamster after a passive outward journey. AnimalBehaviour, 34, 696–715.

Fisher, N. (1993). Statistical analysis of circular data. Cambridge U.K.:Cambridge University Press.

Frohardt, R., Marcroft, J., & Taube, J. P. (2002). Do head direction cellsguide spatial navigation in rats?: An electrophysiological investigationin a path integration task. Society for Neuroscience Abstracts, Society ofNeuroscience, Washington, DC.

Frohardt, R. J., Bassett, J. P., & Taube, J. S. (2006). Path integration andlesions within the head direction cell circuit: Comparison between theroles of the anterodorsal thalamus and dorsal tegmental nucleus. Behav-ioral Neuroscience, 120, 135–149.

Gallistel, C. (1990). The organization of learning. Bradford, Cambridge,MA.

Golob, E. J., Stackman, R. W., Wong, A., & Taube, J. S. (2001). On thebehavioral significance of head direction cells: Neural and behavioraldynamics during spatial memory tasks. Behavioral Neuroscience, 115,285–304.

Goodridge, J. P., & Taube, J. S. (1997). Interaction between the postsub-iculum and anterior thalamus in the generation of head direction cellactivity. Journal of Neuroscience, 17, 9315–9330.

Hargreaves, E. L., Yoganarasimha, D., & Knierim, J. J. (2007). Cohesive-ness of spatial and directional representations recorded from neuralensembles in the anterior thalamus, parasubiculum, medial entorhinalcortex, and hippocampus. Hippocampus, 17, 826–841.

Jeffery, K. J., Donnett, J. G., Burgess, N., & O’Keefe, J. M. (1997).Directional control of hippocampal place fields. Experimental BrainResearch, 117, 131–142.

Jeffery, K. J., & Etienne, A. S. (2004). Path integration in mammals.Hippocampus, 14, 180–192.

Jeffery, K. J., Gilbert, A., Burton, S., & Strudwick, A. (2003). Preservedperformance in a hippocampal-dependent spatial task despite completeplace cell remapping. Hippocampus, 13, 175–189.

Knierim, J., Kudrimoti, H., & McNaughton, B. (1998). Interactions be-tween idiothetic cues and external landmarks in the control of place cellsand head direction cells. Journal of Neurophysiology, 80, 425–446.

Lenck-Santini, P.-P., Muller, R. U., Save, E., & Poucet, B. (2002). Rela-tionships between place cell firing fields and navigational decisions byrats. Journal of Neuroscience, 22, 9035–9047.

Leutgeb, S., Ragozzino, K. E., & Mizumori, S. J. (2000). Convergence ofhead direction and place information in the CA1 region of hippocampus.Neuroscience, 100, 11–19.

Maaswinkel, H., & Whishaw, I. Q. (1999). Homing with locale, taxon, anddead reckoning strategies by foraging rats: Sensory hierarchy in spatialnavigation. Behavioural Brain Research, 99, 143–152.

McNaughton, B., Chen, L., & Markus, E. (1991). ‘Dead reckoning,’landmark learning, and the sense of direction: A neurophysiological and

168 BRIEF COMMUNICATIONS

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

computational hypothesis. Journal of Cognitive Neuroscience, 3, 190–202.

McNaughton, B. L., Battaglia, F., Jensen, O., Moser, E. I., & Moser, M.-B.(2006). Path integration and the neural basis of the ’cognitive map’.Nature Reviews Neuroscience, 7, 663–78.

Mittelstaedt, M., & Mittelstaedt, H. (1980). Homing by path integration inthe mammal. Naturwissenschaften, 67, 566–567.

Mizumori, S. J. Y., & Williams, J. (1993). Directionally selective mne-monic properties of neurons in the lateral dorsal nucleus of the thalamusof rats. Journal of Neuroscience, 13, 4015–4028.

Moser, E. I., Kropff, E., & Moser, M.-B. (2008). Place cells, grid cells, andthe brain’s spatial representation system. Annual Review of Neuro-science, 31, 69–89.

Muir, G. M., & Taube, J. S. (2002). The neural correlates of navigation: Dohead direction and place cells guide spatial behavior? Behavioral andCognitive Neuroscience Reviews, 1, 297–317.

Muir, G. M., & Taube, J. S. (2004). Head direction cell activity andbehavior in a navigation task requiring a cognitive mapping strategy.Behavioural Brain Research, 153, 249–253.

O’Keefe, J., & Nadel, L. (1978). The hippocampus as a cognitive map.Oxford: Clarendon Press; New York.

Pearce, J. M., Roberts, A. D., & Good, M. (1998). Hippocampal lesionsdisrupt navigation based on cognitive maps but not heading vectors.Nature, 396, 75–77.

Ranck, J. B., Jr. (2005). Foreword. In S Wiener, and J. Taube (Eds.), Headdirection cells and the mechanisms of spatial orientation, pp. vi–viii.Cambridge, MA: MIT Press.

Redish, A. D. (1999). Beyond the cognitive map: From place cells toepisodic memory. Cambridge, MA: MIT Press.

Schone, H. (1984). Spatial orientation: Spatial control of behavior inanimals and man. Princeton University Press, Princeton, NJ.

Seguinot, V., Maurer, R., & Etienne, A. S. (1993). Dead reckoning in asmall mammal: The evaluation of distance. Journal of ComparativePhysiology, A, 173, 103–113.

Shrager, Y., Kirwan, C. B., & Squire, L. R. (2008). Neural basis of thecognitive map: Path integration does not require hippocampus or ento-

rhinal cortex. Proceedings of the National Academy of Sciences of theUnited States of America, 105, 12034–12038.

Taube, J. S. (2007). The head direction signal: Origins and sensory-motorintegration. Annual Review of Neuroscience, 30, 181–207.

Taube, J. S., & Bassett, J. P. (2003). Persistent neural activity in headdirection cells. Cerebral Cortex, 13, 1162–1172.

Tolman, E., Ritchie, B., & Kalish, D. (1946). Studies in spatial learning. I.Orientation and the short-cut. Journal of Experimental Psychology, 36,13–24.

van der Meer, M. A. A., Knierim, J. J., Yoganarasimha, D., Wood, E. R.,& van Rossum, M. C. W. (2007). Anticipation in the rodent headdirection system can be explained by an interaction of head movementsand vestibular firing properties. Journal of Neurophysiology, 98, 1883–1897.

Vann, S. D. (2005). Transient spatial deficit associated with bilaterallesions of the lateral mammillary nuclei. European Journal of Neuro-science, 21, 820–824.

Wallace, D. G., Gorny, B., & Whishaw, I. Q. (2002). Rats can track odors,other rats, and themselves: Implications for the study of spatial behavior.Behavioural Brain Research, 131, 185–192.

Wallace, D. G., & Whishaw, I. Q. (2003). NMDA lesions of ammon’s hornand the dentate gyrus disrupt the direct and temporally paced homingdisplayed by rats exploring a novel environment: Evidence for a role ofthe hippocampus in dead reckoning. European Journal of Neuroscience,18, 513–523.

Whishaw, I. Q., & Tomie, J. (1997). Piloting and dead reckoning dissoci-ated by fimbria-fornix lesions in a rat food carrying task. BehaviouralBrain Research, 89, 87–97.

Wilton, L. A., Baird, A. L., Muir, J. L., Honey, R. C., & Aggleton, J. P.(2001). Loss of the thalamic nuclei for “head direction” impairs perfor-mance on spatial memory tasks in rats. Behavioral Neuroscience, 115,861–869.

Received August 18, 2009Revision received November 30, 2009

Accepted November 30, 2009 �

169BRIEF COMMUNICATIONS

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.