Dorsal striatal head direction and hippocampal place representations during spatial navigation

Abstract

Several theories of basal ganglia function describe a striatal contribution to learning that is independent of hippocampal function. This study examined the question of whether the striatum should be regarded as functioning independently of or acting in concert with limbic structures. Dorsal striatal head direction cells and hippocampal place cells were recorded in parallel while rats performed a hippocampal-dependent radial maze task. Changes in the directional preference of head direction cells and the location of place fields were compared following alterations of the sensory environment. When familiar visual cues were presented in new spatial arrangements, or when new visual cues were placed in a familiar environment, rotations of directional preferences were consistent with the mean place-field response. When familiar visual and nonvisual cues were presented in conflict, or when rats were exposed to novel environments, the responses of the two cell types were inconsistent relative to each other. This pattern suggests that current perceptions and expectations of familiar spatial contexts may dynamically modulate the relationship between hippocampus and dorsal striatum.

Full text

Exp Brain Res (2001) 139:372–376
DOI 10.1007/s002210100795
Abstract Several theories of basal ganglia function de-
scribe a striatal contribution to learning that is indepen-
dent of hippocampal function. This study examined the
question of whether the striatum should be regarded as
functioning independently of or acting in concert with
limbic structures. Dorsal striatal head direction cells and
hippocampal place cells were recorded in parallel while
rats performed a hippocampal-dependent radial maze
task. Changes in the directional preference of head direc-
tion cells and the location of place fields were compared
following alterations of the sensory environment. When
familiar visual cues were presented in new spatial ar-
rangements, or when new visual cues were placed in a
familiar environment, rotations of directional preferenc-
es were consistent with the mean place-field response.
When familiar visual and nonvisual cues were presented
in conflict, or when rats were exposed to novel environ-
ments, the responses of the two cell types were inconsis-
tent relative to each other. This pattern suggests that cur-
rent perceptions and expectations of familiar spatial con-
texts may dynamically modulate the relationship be-
tween hippocampus and dorsal striatum.
Keywords Basal ganglia · Limbic system · Navigation ·
Rat
Introduction
Recent evidence suggests that multiple brain systems
contribute to spatial learning. Spatial deficits result from
damage to cortical and subcortical structures including
hippocampus (HPC; Morris et al. 1982) and dorsal stria-
tum (DS; Devan et al. 1999). Also, both HPC and DS
neurons exhibit spatial codes such as location-specific
firing (place fields) or head direction (HD) firing that is
consistent with an animal’s heading direction indepen-
dent of its location (O’Keefe and Dostrovsky 1971;
Wiener 1993; Mizumori et al. 1999, 2000; Leutgeb et al.
2000). Other data suggest that HPC and DS make inde-
pendent contributions to learning, with only HPC being
selectively involved in spatial processing (McDonald
and White 1993; Packard and McGaugh 1996). Here we
tested whether the HPC and DS should be considered as
functioning independently during spatial learning by re-
cording simultaneously DS HD cells and HPC place
cells as rats performed a spatial maze task. Unit respons-
es to various environmental manipulations were com-
pared.
Methods
All methods described here were approved by the University of
Utah IACUC. Male rats were individually housed on a 12-h light/
dark cycle and reduced to 80% of ad libitum weights. Behavioral
testing and unit recording took place within a black-curtained,
square environment (158 cm×158 cm×305 cm) while rats per-
formed on an eight-arm, remote-controlled radial maze (Mizumori
and Williams 1993). The maze arms (58 cm×5.5 cm) radiated
from a round central platform (19-cm diameter). Symmetrical
lighting was provided within the curtained arena. For standard
testing situations, visual cues were attached to the curtains. The
computer and recording equipment were located in a room adja-
cent to the maze room.
Six rats were recorded during asymptotic performance levels
in which a different sequence of maze arms had to be recalled
each trial (Mizumori and Williams 1993). The first four arms in
the sequence was randomly selected and individually presented to
the rat (forced choice procedure). All eight arms were then made
available to the rat. Re-entries into maze arms already entered on
that trial constituted working memory errors.
When rats completed eight trials within 1 h, they were anesthe-
tized for stereotaxic surgery. Four stereotrodes, two per hemi-
sphere, were implanted (DS: AP +0.2–1.2 mm, L ±1.5 mm, DV
2.0 mm; HPC: AP –3.5–4.5 mm, L ±2.5 mm, DV 1.5 mm). In ac-
cordance with previously described methodologies (Leutgeb and
Mizumori 1999; Mizumori and Williams 1993), neural activity de-
K.E. Ragozzino · S. Leutgeb
Psychology Department and Program in Neuroscience,
390E. 1530S., Rm 502, University of Utah, Salt Lake City,
Utah 84112, USA
S.J.Y. Mizumori (
✉
)
Psychology Department, Box 351525, University of Washington,
Seattle, WA 98195, USA
e-mail: mizumori@u.washington.edu
Tel.: +1-206-5432699, Fax: +1-206-6853157
RESEARCH NOTE
Katharine E. Ragozzino · Stefan Leutgeb
Sheri J. Y. Mizumori
Dorsal striatal head direction and hippocampal place representations
during spatial navigation
Received: 30 June 2000 / Accepted: 9 May 2001 / Published online: 28 June 2001
© Springer-Verlag 2001

373
tected on the eight channels was recorded simultaneously and in-
dependently with a Datawave Workstation. Each spike occupied a
distinct location in a parametric space that took into account nu-
merous waveform properties. Spikes from a given cell tend to
form a cluster of spikes. These clusters were matched across days
on the basis of similarity of cluster boundaries and waveform
characteristics. HD cells were recorded using either single-diode
or double-diode tracking. Consistent with previous comparisons of
HD responses recorded with either single- or double-diode meth-
ods (Leutgeb et al. 2000), changes in directional preferences fol-
lowing experimental manipulations were identical when recorded
with either tracking method. Note, however, a possible difference
in the maximum amplitude of response in the preferred direction.
This apparent difference may be due to the different resolutions of
directional movement offered by the two tracking methods. All re-
cording sites were verified histologically.
The following manipulations were performed to test (a) the rel-
ative contribution of visual and nonvisual information to the spa-
tial correlates (lights off manipulation), (b) responses to cue con-
flict situations (maze and cue rotation manipulations), and (c) sen-
sitivity to changes in visual cue reliability (novel/altered environ-
ment conditions).
Lights off
Rats performed 5 trials with the lights on (2-min intertrial inter-
val), followed by 5 trials with the lights turned off, and 5 trials
with the lights restored (light-dark-light, or LDL condition). Rats
were not removed from the maze while lights were turned off and
on.
Maze rotation
Rats were brought into the maze room in darkness with the maze
rotated 45° or 90° counterclockwise (CCW). They performed
5 trials in darkness, then 5 trials with the lights on. Animals ini-
tially entered the test area through one of four entrances (randomly
determined). The maze remained rotated throughout the test.
Cue rotation
Rats first performed 5 trials with the visual cues in standard con-
figurations and locations (baseline condition), 5 trials with all cues
rotated 180°, then 5 trials with cues restored. The animals were re-
moved from the room while the cues were rotated.
Novel room/altered familiar room
HD and place cells were recorded in the familiar environment,
then recorded (1) when rats performed in a structurally similar but
different maze room that contained novel visual cues (novel envi-
ronment condition), or (2) following exposure to novel visual cues
or novel visual cue arrangement in the familiar environment (al-
tered familiar environment condition). For the altered familiar en-
vironment condition, the curtains surrounding the maze were
raised (providing rats with a larger visual environment and new
cues), or the distal cues were scrambled (providing new spatial re-
lationships between familiar cues). In both conditions, the rat was
removed temporarily from the maze room during the cue manipu-
lations.
Behavioral correlations
For spatial memory recording sessions with unique cell pairs, we
tested for correlations (during the manipulated conditions) be-
tween HD tuning and errors, and between place cell specificity/
reliability and errors. Location specificity reflected the relative
difference between in-field and out-field firing rates, and reliabil-
ity measures reflected the proportion of trials in which the maxi-
mum firing occurred within the defined place field (McNaughton
et al. 1983). HD cell directional specificity (tuning) was deter-
mined by first calculating the firing rates as the rat moved in the
preferred direction, then dividing by the mean rate as the rat
moved in the remaining seven directions offered by the maze
arms. For sessions recorded with a single diode, only outbound
rates were used because, as verified in offline analysis, they re-
present the longest uninterrupted bout of the same behavior in
each of the eight directions.
Spatial correlation analysis
To compare the responses of HD and place cells, we calculated the
angular deviation between baseline and each manipulated condi-
tion. Since space was sampled in 45° steps, we restricted the reso-
lution of the analysis to these increments. For HD cells, the direc-
tional tuning was estimated by first calculating the discharge rate
for each of the eight directions in baseline and manipulated condi-
tions. These conditions were compared by calculating pairwise
correlation coefficients while shifting the maze orientations in 45°
steps with respect to each other. The angular deviation that corre-
sponded to the maximum pairwise correlation coefficient was de-
fined as the characteristic response for each recording session. The
angular deviation of HPC place fields was measured with a similar
algorithm but with a polar coordinate system. When more than
one HD cell or more than one place cell was recorded, correlation
coefficients were averaged for each cell type.
Results
Thirty-four recording sessions are described in which
one or two DS HD cells and one to five hippocampal
place cells were recorded simultaneously. Seven sessions
included two simultaneously recorded HD cells and
15 sessions included more than one place cell. A total of
11 HD cells were recorded with 20 place cells. Within
each manipulation condition, the recording sessions were
of unique cell pairs. Many of the same cell pairs were
tested across several manipulations. The following de-
scription of changes in the directional preferences of HD
cells reflects a change in the preferred direction, and not
elevated firing during movements in many directions.
Altered illumination
Thirteen HPC place cells and nine DS HD cells were
recorded in eight LDL recording sessions. In seven ses-
sions, HD and place cell responses were similar in that
they were not significantly altered by changes in room
illumination (Fig. 1A). The persistence of spatial corre-
lates in darkness was observed when rats performed the
spatial memory task in a familiar environment, when rats
had minimal exposure (less than 5 days) to the environ-
ment, and when rats performed in novel or altered famil-
iar environments. In one session, both HD (n=1) and
place cell (n=1) changed their response properties in
darkness.

374
Maze rotation
Twelve HPC place cells and nine DS HD cells were re-
corded in seven maze rotation sessions (Fig. 1B). A vari-
ety of responses were observed. In some cases, place and
HD preferences were found to both rotate either along
with or independent of the maze (four sessions; n=1 or 2
HD cells/session; n=1–5 place cells/session). In other
sessions either place fields or HD preferences rotated
while the other cell response type did not (three sessions;
n=1 HD cell/session; n=1 place cell/session).
Cue rotation of 180°
Following 180° visual cue rotations (seven sessions in-
volving 7 HD cells and 15 place cells), four patterns of
results were observed (see Fig. 1B): (1) both place (n=1)
and HD cell (n=1) rotated their spatial preferences by
180° (one session); (2) neither cell type rotated with the
cues (one session; n=1 HD cell; n=3 place cells); (3) HD
cells rotated with the cues while place cells ceased firing
or fired in new locations (two sessions; n=1 HD cell/ses-
sion; n=1 to 3 place cells/session); (4) HD cells did not
rotate, and place cells ceased firing or fired in new loca-
tions (two sessions; n=1 HD cell/session; n=1 place
cell/session; also see Fig. 2B); or (5) the HD cell (n=1)
rotated its directional preference slightly while the place
field (n=1) did not change (one session).
Novel environment
In the novel environment (Fig. 1B), all eight HD cells
tested (five recording sessions) rotated 180° in absolute
compass direction relative to their orientation in the fa-
miliar environment (e.g., if the cell fired magnetic north
Fig. 1A, B Scattergrams comparing the angular deviation of hip-
pocampus (HPC) place fields or head direction (HD) preferences
(relative to baseline trials) following different experimental ma-
nipulations. Each point is the mean response of all place fields or
all HD preferences recorded in a single recording session. A A
significant correlation was found between dorsal striatal HD and
hippocampal place cell responses following changes in the con-
stellation or spatial arrangement of distal visual cues in an other-
wise familiar environment, indicating that both cell types tended
to respond in parallel. In addition, most HD and place cells re-
mained unchanged during the light-dark-light (LDL) manipula-
tion. In one session, place fields reorganized during darkness, then
returned to baseline locations after the restoration of lights. B Ma-
nipulations that presented conflicts with familiar cues or a novel
environment resulted in uncorrelated responses by HD and place
cells
Fig. 2A, B Directional tuning curves for HD cells and spatial dis-
tribution plots of simultaneously recorded place fields. A Both
correlate types showed a similar degree of change following a cue
scramble manipulation. Peak firing rate of the HD cell and the
location of the place field shifted by about 90°. (Circles identify
locations of high firing; overlapping circles occur with repeated
entries into the place field; visited locations are shown by dots)
B When familiar cues were rotated, HD cells tended to respond
independently of place fields. The tuning curves illustrate similar
shifts in directional preference regardless of whether the data
were subject to single (solid lines) or double (dashed lines) diode
analysis

in the familiar environment, it would fire due south in
the novel environment). For novel and familiar recording
rooms, the maze is located in the same place relative to
an adjacent computer room and the room entrance. Im-
portantly, however, in the familiar room, the room en-
trance and computer room are north of the maze, while
in the novel room they are located south of the maze.
Rats were not disoriented upon entering the room. Thus,
the HD cells could have relied on either static back-
ground cues (e.g., auditory cues) or self-motion cues.
HPC place cell responses (n=10) to a novel room
were more diverse. For example, in one session, a place
cell (n=1) from one animal rotated 180° along with the
HD cells’ (n=2) directional preferences. However, for a
second animal in which four place cells were recorded
simultaneously, one place cell developed a different field
and three place fields rotated 180° along with the HD
cell’s (n=1) preferred direction. Two of the rotated place
fields remained as such over days. One other place field
that initially rotated with the HD preference developed a
different and distinct field on the second exposure to the
novel environment even though the HD cell’s preferred
directions remained the same. This mixture of rotating
and nonrotating place fields is reflected in mean rotation
values other than 180°. In two other sessions, the single
place cell recorded did not change in concordance with
the single HD cell recorded. Thus, HD cell responses
were not consistent with all place cell responses.
Altered environment
Changing either the spatial relationship of familiar cues
or adding new visual cues in an otherwise familiar room
resulted in similar rotations of place fields and HD pref-
erences (Fig. 1A; r=0.98, P<0.001). For two sessions
tested in the cue scramble condition and one session in
which all visual cues were removed, the four HD and
nine place cells tested showed comparable degrees of
change in terms of displacement of HD preferences or
place field location. The changed location of the place
fields or the change in directional preferences were not
associated with the new location of specific cues. The
spatial firing returned to their baseline correlate when
the distal visual cues were repositioned to their familiar
configuration. When the familiar test room was altered
by raising the black curtains surrounding the maze (four
sessions from two rats), again the place fields (n=14) and
HD cells’ (n=5) preferred direction rotated in alignment
with each other.
Correspondence between spatial firing
and choice accuracy
A significant correlation between the number of errors
and HD cell directionality was found (r=–0.277,
P<0.025), indicating a tendency for errors to increase
when the tuning of HD cells was less specific. The corre-
lation for place-field specificity/reliability and errors was
not significant, nor was there a statistical correlation be-
tween choice accuracy and times when HD preferences
and place fields responded similarly or differently. As
described in previous investigations of rat performance
in darkness (Brown and Bing 1997; Save 1997), rats in
the present study were relatively undisturbed in terms of
choice accuracy during dark trials.
Discussion
Spatial representations in DS and HPC responded simi-
larly during the LDL and the altered familiar environ-
ment conditions. In these situations, visual aspects of a
familiar environment underwent such dramatic changes
that previously reliable visual information became unre-
liable. That both HD preferences and place fields re-
mained aligned when visual cues became unreliable sug-
gests that these cells were dependent upon a common
nonvisual spatial coordinate system. That is, they may
have become coupled in the absence of external visual
cues that served to align the spatial correlates in light
conditions. One interpretation of this pattern of results is
that current perceptions and expectations of familiar spa-
tial contexts modulate the relationship between hippo-
campal and dorsal striatal representations.
One of two conditions that frequently led to different
patterns of change for HD and place cells presented a
cue conflict situation (by cue or maze rotation) in which
the expected relationship between extramaze and
intramaze sensory conditions was disrupted. Consistent
with the literature on place field reorganization phenom-
ena, place fields reorganized during cue conflict situa-
tions. The proportion of place cells that rotated with the
visual cues was lower than previously reported (Muller
and Kubie 1987; Bostock et al. 1991), perhaps due to a
difference in the behavioral history of the animals. The
cue rotation sessions of this study occurred in well-
trained animals and after other visual manipulations, a
situation that may result in less cue control over place
fields (Knierim et al. 1995). HD cells, on the other hand,
tended to either not change or follow salient landmarks
even when recorded simultaneously with place cells.
The second condition that led to different responses
for HD and place cells was exposure to a novel room
with novel cues. All HD cells tested rotated their direc-
tional preferences by 180°, while the place fields reorga-
nized in unpredictable ways. HD cells probably used
common static background cues to align directional pref-
erences. In contrast place cells appeared to have greater
reliance on unique distal cues, showing near complete
reorganization when animals were placed in a novel en-
vironment (Bostock et al. 1991; Knierim et al. 1995).
The degree of conditional alignment of HD and place
cells observed in this study can be compared with the
coupling reported for HD cells of the anterior nucleus of
the thalamus (ATN) and HPC place cells (Knierim et al.
1995). Differences in the behavioral testing history of
375

Devan B, McDonald R, White N (1999) Parallel information pro-
cessing in the dorsal striatum: relation to hippocampal func-
tion. J Neurosci 19:2789–2798
Knierim J, Kudrimoti H, McNaughton BL (1995) Place cells, head
direction cells, and the learning of landmark stability. J Neuro-
sci 15:1648–1659
Knowlton B, Mangels JA, Squire LR (1996) A neostriatal habit
learning system in humans. Science 273:1399–1402
Leutgeb S, Mizumori SJY (1999) Excitotoxic septal lesions result
in spatial memory deficits and altered flexibility of hippocam-
pal single-unit representations. J Neurosci 19:6661–6672
Leutgeb S, Ragozzino KE, Mizumori SJY (2000) Convergence of
head direction and place information in the CA1 region of hip-
pocampus. Neuroscience 100:11–19
McDonald R, White N (1993) A triple dissociation of memory
systems: hippocampus, amygdala and dorsal striatum. Behav
Neurosci 107:3–22
McNaughton BL, Barnes CA, O’Keefe J (1983) The contributions
of position, direction and velocity to single unit activity in the
hippocampus of freely moving rat. Exp Brain Res 52:41–49
Mizumori SJY, Williams JD (1993) Directionally selective mne-
monic properties of neurons in the lateral dorsal nucleus of the
thalamus of rats. J Neurosci 13:4015–4028
Mizumori SJY, Pratt WE, Ragozzino KE (1999) Function of the
nucleus accumbens within the context of the larger striatal
system. Psychobiology 27:214–224
Mizumori SJY, Ragozzino KE, Cooper BG (2000) Location and
head direction representation in the dorsal striatum of rats.
Psychobiology 28:441–462
Morris RGM, Garrud P, Rawlings J, O’Keefe J (1982) Place navi-
gation impaired in rats with hippocampal lesions. Nature
297:681–683
Muller RU, Kubie JL (1987) The effects of changes in the envi-
ronment on the spatial firing of hippocampal complex-spike
cells. J Neurosci 7:1951–1968
O’Keefe J, Dostrovsky J (1971) The hippocampus as a spatial
map. Preliminary evidence from unit activity in the freely-
moving rat. Brain Res 34:171–175
Packard M, McGaugh JL (1996) Inactivation of hippocampus or
caudate nucleus with lidocaine differentially affects expression
of place and response learning. Neurobiol Learn Mem 65:
65–72
Save E (1997) The contribution of visual and inertial mechanisms
to navigation in total darkness. Anim Learn Behav 25:
324–334
Taube JS (1998) Head direction cells recorded and the neurophysi-
ological basis for a sense of direction. Prog Neurobiol 55:
225–256
Wiener S (1993) Spatial and behavioral correlates of striatal neu-
rons in rats performing a self-initiated navigation task. J. Neu-
rosci 13:3802–3817
376
ATN and DS HD cells may give the appearance of a
functional distinction when none exist. That is, the fact
that this study found a greater proportion of HD and
place cell pairs that did not respond in a similar fashion
may be due to the more drastic environmental manipula-
tions used in the present experiment. In the future, simul-
taneous recordings of DS and ATN HD cells are needed
to resolve this issue.
If it is the case that there is no significant difference
in the response of ATN and DS HD cells, an interesting
question to ask is why a similar neural code is needed for
two brain structures that are considered to make different
contributions to learning. The ATN seems important for
understanding one’s directional sense during navigation
(Taube 1998), and the DS is typically considered im-
portant for mediating procedural or habit learning
(McDonald and White 1993; Knowlton et al 1996;
Packard and McGaugh 1996). It is possible that there are
multiple component processes required for accurate per-
formance of a working memory task on a radial maze.
As animals solve the task, knowledge about the spatial
context must be integrated with knowledge about the
current and expected reinforcement conditions and re-
sponse options. Perhaps, HD signals are useful in multi-
ple neural systems because they provide a common ref-
erence frame within which to interpret different kinds of
information. That is, depending on one’s orientation in
an environment, there may be a different meaning to cur-
rent spatial context, reinforcement, or response option
information. In this way, both HPC and DS may make
complementary, not completely independent, contribu-
tions to learning.
Acknowledgements This work was supported by NIH grant
MH58755. We thank Brent Cooper and Alex Guazzelli for helpful
discussions.
References
Bostock E, Muller RU, Kubie JL (1991) Experience-dependent
modifications of hippocampal place cell firing. Hippocampus
1:193–206
Brown MF, Bing MN (1997) In the dark: spatial choice when ac-
cess to spatial cues is restricted. Anim Learn Behav 25:21–30