J Neurophysiol 113: 740 –753, 2015.
First published November 5, 2014; doi:10.1152/jn.00114.2014.
Neurocognitive stages of spatial cognitive mapping measured during free
exploration of a large-scale virtual environment
Markus Plank,1 Joseph Snider,1 Erik Kaestner,2 Eric Halgren,2,3* and Howard Poizner1,2*
1
Institute for Neural Computation, University of California, San Diego, La Jolla, California; 2Interdepartmental Neuroscience
Program, University of California, San Diego, La Jolla, California; and 3Departments of Radiology, Neurosciences, and
Psychiatry, University of California, San Diego, La Jolla, California
Submitted 10 February 2014; accepted in final form 2 November 2014
spatial information; temporal cortex; parietal cortex; frontal cortex;
event-related potential; N400; processing negativity; theta; cognitive
map
AS HUMANS AND OTHER ANIMALS move through their surroundings, they integrate multisensory information from visual,
vestibular, and proprioceptive cues to form inner representations of spatial environments (Chance et al. 1998; Klatzky
1998). These representations organize our behavior and form
the core context of our subjective experience (O’Keefe and
Nadel 1978). Neural firing correlates of spatial representations
have been extensively studied in the rodent hippocampal formation as the animals move through environments. These
neural correlates take the form of place (O’Keefe et al. 1998),
grid (Hafting et al. 2005), boundary (Solstad et al. 2008), and
head direction cells (Cho and Sharp 2001)—all of which
stabilize rapidly in novel environments (Hafting et al. 2005)
and are sensitive to alterations of the environment (Muller et al.
1987; Smith and Mizumori 2006). Theta rhythms organize this
* E. Halgren and H. Poizner contributed equally to this work.
Address for reprint requests and other correspondence: H. Poizner, Inst. for
Neural Computation, Univ. of California, San Diego, 9500 Gilman Dr.-0523,
La Jolla, CA 92093-0523 (e-mail: [email protected]).
740
spatially related cell firing, gating memory encoding and retrieval and synchronizing activity across multiple cortical areas, including parietal and frontal cortices (Hasselmo and Stern
2014; Siapas et al. 2005).
Consistent with the central role of space in mental experience, O’Keefe and Nadel (1978) interpreted hippocampal place
cells, and the entirety of hippocampal recording and lesion
studies, as implementing a “cognitive map.” To accommodate
the already large literature demonstrating nonspatial memory
deficits after medial temporal lesions in humans, they posited
that the hippocampus in the dominant hemisphere supported a
verbal cognitive map to complement the spatial cognitive map
supported by the nondominant hemisphere in humans and both
hemispheres in rodents.
In subsequent years, there have been many studies demonstrating the neurocognitive stages that integrate words into the
dynamic cognitive context. The most prominent stage is the
N400, commonly recorded in the scalp EEG but also apparent
in MEG (Halgren et al. 2002) and intracranial local field
potentials (Nobre et al. 1994; Smith et al. 1986) and correlated
with fMRI (Dale et al. 2000) and unit activity (Heit et al.
1988). The N400 is evoked by words and other semantic
stimuli and is smaller when the cognitive context allows the
word to be integrated more easily (Brown and Hagoort 1993;
Halgren 1990; Holcomb et al. 2002; Kutas and Federmeier
2011). The responses to words can also be modulated at earlier
latencies, ⬃200 ms, in a less contextual fashion, by their
lexical frequency (Sahin et al. 2009) or broad semantic category (Chan et al. 2011; Pulvermuller et al. 1999). The N400 is
often followed by the P600, especially when initial associations yield erroneous conclusions (Kuperberg 2007) such as
puns (Marinkovic et al. 2011), and/or by theta activity. Unlike
the N400, theta is larger to actual words and has been interpreted as reflecting sustained memory retrieval processes
(Marinkovic et al. 2012). The N200/N400 to words are generated mainly in the left antero-ventral temporal, temporo-parietal, and postero-ventral prefrontal cortices (Halgren et al.
1994a, 1994b, 2002; Nobre et al. 1994; Smith et al. 1986). The
P600 and theta appear to be generated in classical language
areas, as well as more dorsolateral prefrontal, parietal, and
especially cingulate cortices (Wang et al. 2005). In particular,
the medial temporal structures in and around the hippocampal
formation show strong activation during the N400 (Nobre et al.
1994; Smith et al. 1986), supporting the prediction of O’Keefe
and Nadel (1978) that these areas implement a verbal cognitive
map, or semantic hub for the cortex (Patterson et al. 2007).
0022-3077/15 Copyright © 2015 the American Physiological Society
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.5 on June 18, 2017
Plank M, Snider J, Kaestner E, Halgren E, Poizner H. Neurocognitive
stages of spatial cognitive mapping measured during free exploration of a
large-scale virtual environment. J Neurophysiol 113: 740–753, 2015. First
published November 5, 2014; doi:10.1152/jn.00114.2014.—Using a novel,
fully mobile virtual reality paradigm, we investigated the EEG correlates of spatial representations formed during unsupervised exploration. On day 1, subjects implicitly learned the location of 39 objects
by exploring a room and popping bubbles that hid the objects. On day
2, they again popped bubbles in the same environment. In most cases,
the objects hidden underneath the bubbles were in the same place as
on day 1. However, a varying third of them were misplaced in each
block. Subjects indicated their certainty that the object was in the
same location as the day before. Compared with bubble pops revealing correctly placed objects, bubble pops revealing misplaced objects
evoked a decreased negativity starting at 145 ms, with scalp topography consistent with generation in medial parietal cortex. There was
also an increased negativity starting at 515 ms to misplaced objects,
with scalp topography consistent with generation in inferior temporal
cortex. Additionally, misplaced objects elicited an increase in frontal
midline theta power. These findings suggest that the successive
neurocognitive stages of processing allocentric space may include an
initial template matching, integration of the object within its spatial
cognitive map, and memory recall, analogous to the processing
negativity N400 and theta that support verbal cognitive maps in
humans.
EEG CORRELATES OF SPATIAL COGNITIVE MAPPING
MATERIALS AND METHODS
Subjects
Twelve young, healthy adults without history of psychiatric or
neurological disorder participated in the experiment [3 women, 9 men;
age ⫽ 24.8 ⫾ 4 yr (mean ⫾ SD)]. All participants were right-handed,
as measured with the Edinburgh Handedness Inventory (Oldfield
1971). Participation was reimbursed at a rate of $20/h. The study was
approved by the UCSD Human Research Committee and was conducted according to the principles expressed in the Declaration of
Helsinki. Written informed consent was obtained from all participants.
A4: EEG
Experimental Paradigm
Subjects actively explored a novel VE rendered on a wide-field of
view head-mounted display (HMD). Subjects physically moved at
will in a large-scale, richly textured virtual room (⬃4.0 ⫻ 5.0 m)
containing numerous objects resting on shelves, tables, and the floor
(see Fig. 1). This virtual room, located on an “aircraft carrier,” was the
same size as the real-world space in which subjects were walking.
The experiment comprised two sessions lasting 2 h each, which
were conducted on two subsequent days. On day 1, participants first
explored the room for 10 min with no instruction except “explore the
environment.” Then, during five subsequent blocks, opaque virtual
bubbles were placed around 39 different objects in the room. The
subjects walked up to one bubble at a time (indicated by turning
green) and popped it by touching it with their hand, thereby uncovering the object underneath (the objects had been present during the
initial exploration but not obscured by the bubbles). The popping of
the virtual bubble was used as a time-locking event for the eventrelated potential (ERP) analysis and was logged into an event file
together with a time stamp that allowed for off-line synchronization
and alignment with the EEG system. As a cover task, subjects
indicated their interest in the object with a virtual slider. A block
ended when all the objects were uncovered and visible. For each of
these five “interest” blocks, the order of bubble popping was varied
pseudorandomly. However, the objects remained the same, thus ensuring both multiple paths through the VE and opportunity for
unsupervised learning of the object locations.
While the first day was dedicated to unsupervised spatial exploration, the second day involved testing subjects’ memory of the environment, in particular whether objects were still placed in their
expected spatial locations. Since subjects were completely naive about
the memory aspect of the experiment when they performed the
interest rating task on day 1, the knowledge of the environment on day
2 must have resulted from unsupervised learning. After an overnight’s
sleep to allow for memory consolidation, on day 2 the subjects entered
the same virtual room. This time, the locations of a subset of 13 of the
39 objects were randomly interchanged from their positions on day 1.
A1: bi
bird’s
s eye
e view
vi w of simul
simulated
ed e
envi
vironme
onment
A2:
A3:
B1
-225 ms
B2
0 ms
B3
1000 - 1250 ms
B4
2200 - 2800 ms
J Neurophysiol • doi:10.1152/jn.00114.2014 • www.jn.org
Fig. 1. A: experimental environment. Participants freely explored a virtual storage room
(A1; bird’s eye view, never shown to participants), matched in size (4 ⫻ 5 m) to the
physical room in which the participant was
physically present (A2). The virtual environment was presented via a head-mounted display (A3; ego view). During the session, the
continuous EEG of 72 active electrodes was
collected (A4; 2-dimensional voltage map). B:
trial sequence: the experimental task was to
walk toward a green bubble (B1), which disappeared upon being touched with the right
hand (B2), also accompanied by a popping
sound, thereby revealing the hidden object
and providing the time-locking event for neurophysiological analyses. After a randomized
time interval of 1,000 –1,250 ms, a virtual
slider bar appeared in front of the subject (B3).
On day 1, subjects used the slider bar to rate
“how interesting” the object was. On day 2,
subjects again approached and popped green
bubbles but used the slider to indicate whether
the revealed object was correctly placed or misplaced. After adjusting the slider (no time limit),
participants lifted their hand above the slider
(B4) to confirm their adjustment, which activated the next green bubble. The time interval
between 225 and 25 ms prior to bubble pop
(starting at B1) was used as baseline period for
the event-related potential (ERP) analysis.
Downloaded from http://jn.physiology.org/ by 10.220.33.5 on June 18, 2017
Recording studies of spatial cognitive maps in humans under
realistic conditions have been rare, largely because of the head
and body restraints usually required in fMRI, MEG, or even
EEG recordings. Because spatial awareness requires integration of visual, vestibular, and proprioceptive cues during locomotion (Klatzky 1998), investigating the brain processes underlying spatial learning is best done during fully mobile
exploration in naturalistic settings. We have resolved the recording problem and have examined brain activity in fully
mobile human participants interacting in an immersive largescale virtual environment (VE). On day 1, EEG was recorded
continuously as participants actively explored a novel environment containing 39 distinctive objects at fixed locations. On
day 2, subjects were confronted with objects that were in the
expected location or were misplaced. We hypothesized that
comparing the response to expected versus misplaced objects
would reveal neurocognitive processing stages for spatial cognitive mapping, namely, an early first-pass processing of allocentric space followed by an integration of the object within its
spatial cognitive map that would be associated with an N400like processing negativity.
741
742
EEG CORRELATES OF SPATIAL COGNITIVE MAPPING
As on day 1, subjects walked to a bubble covering an object when
it turned green and touched the bubble, exposing the object. Subjects
then indicated how certain they were that the object was the same one
that was in that location on day 1, again by adjusting the virtual slider.
This process was repeated over five blocks of trials, with each block
lasting 5– 8 min. A different set of 13 objects were shuffled in each
block.
Details on the VE
EEG Acquisition and Analysis
EEG data were recorded from 64 Biosemi Active electrodes placed
in a flexible cap according to the extended International 10 –20
System (American Electroencephalographic Society 1991) with a
sampling rate of 1,024 Hz. Eight external electromyographic (EMG)
electrodes were added to help identify muscular artifacts and were
placed as follows. Four electrooculogram (EOG) electrodes were
J Neurophysiol • doi:10.1152/jn.00114.2014 • www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.5 on June 18, 2017
The hardware generating the VE comprised a 24-camera PhaseSpace Impulse active infrared-emitting diode (IRED) motion capture
system (http://www.phasespace.com). With this system, the threedimensional (3D) positions of the right hand, head, and torso were
recorded. In addition, an Intersense (www.intersense.com) InertiaCube3 orientation sensor was mounted on the HMD in order to collect
more accurate orientation data and smooth out shifts of the display
due to head rotations. Auditory stimulation was provided by means of
a 20-speaker Ambisonic Auralizer Sound System (www.worldviz.
com). The VE was rendered on a Sensics (www.sensics.com) xSight
6123 HMD, which provided a panoramic, stereo view of the VE (120°
horizontal and 45° vertical field of view). The VE was programmed in
Vizard 3.0 (www.worldviz.com) with a total end-to-end latency of
⬍40 ms (time from real space movement to update of the visual scene
in the HMD) and spatial precision on the order of millimeters (see
Snider et al. 2013a for a detailed description of the system). There was
an initial constant latency of 15 ms due to the motion capture system,
which is well under latencies that would produce a noticeable lag
between actual and rendered scenes in the HMD. Additionally, there
was a constant 25-ms latency for rendering the visual scene (Snider et
al. 2013a), which was subtracted from the latencies of all events.
The dimensions of the virtual room matched those of the real room
in which the experiment took place, so that the gain between movements in the real and virtual rooms was 1:1. The virtual room
resembled a storage room with a large empty space in the center
surrounded by shelves and tables on which a variety of objects rested.
The objects were richly detailed and textured 3D renderings, some of
which were created from photographs taken on the U.S.S. Midway
naval museum, San Diego, CA (http://www.midway.org). All objects were appropriate to the context of an aircraft carrier setting. Exits
were obstructed with virtual obstacles except for an open virtual door
at which all blocks began. Subjects could look through the door to see
a virtual flight deck but were instructed not to pass beyond the limits
of the room. The position of the subject’s head, as tracked by the
IRED system, and its orientation from the inertial sensor were directly
fed into the VE and refreshed the rendering of the main view in real
time and with a 1:1 mapping of real onto virtual space. The IRED and
inertial sensors were combined to maximize the responsiveness and
stability of the system for improved immersion. The subject’s right
hand was tracked and rendered as a visible, 5-cm-diameter orange ball
that the subjects used to indicate interest (day 1) or certainty (day 2)
on a continuously variable slider that appeared in the VE after bubble
popping. All of the interactions took place such that touching any
object in the real environment was avoided, therefore minimizing any
potential haptic-vision mismatch. Further details on the environment
can be found in Snider et al. (2013a), which analyzed the periods
when the subjects were walking between objects. Here we analyze
events that are time-locked to the bubble pops.
mounted on the supra- and infraorbital ridges of the right eye as well
as lateral to the outer canthi of right and left eyes. Two electrodes
were further attached to the right and left neck at the height of the 7th
cervical vertebra, monitoring activity of neck muscles, particularly the
trapezius. A final two electrodes were mounted on the left and right
mastoids. For all participants, electrode locations were digitized with
a FASTRAK system (Polhemus, Colchester, VT) in combination with
the Locator Suite (Source Signal Imaging, San Diego, CA).
The participants wore a lightweight backpack that held the batterydriven amplifiers. Active EEG electrodes were connected to the
amplifiers via short ribbon cables, comparable to stationary EEG
setups. Amplifiers were connected to the recording computer via a
long fiber-optic cable, but the already-amplified EEG signals were not
contaminated by electrical or mechanical artifacts on this cable.
Data processing was accomplished with custom MATLAB scripts
(The MathWorks, Natick, MA) for EEGLAB (Delorme and Makeig
2004) and the Mass Univariate Statistics Toolbox (Groppe et al.
2011). For visualization purposes, BrainVision Analyzer 2 (Brain
Products, Gilching, Germany) and EEGLAB were employed. In
EEGLAB, data were rereferenced off-line to average mastoids. Data
were high-pass filtered at 1 Hz to remove offset and trend, low-pass
filtered at 55 Hz, and epoched time-locked to bubble pop (epoch
windows covered the time interval between ⫺2,025 and ⫹1,975 ms).
Epoched EEG data were visually inspected for artifactual activity
arising from motor processes, punctual bursts of EMG activity, and
other nonstereotypical processes, and epochs containing artifacts were
excluded from further analyses. On average, 82% of the misplaced
and 78% of the correctly placed trials survived the artifact rejection
procedure (difference not significant, t-testdf ⫽ 11 ⫽ 0.8; P ⫽ 0.25).
After removal of data sections containing artifacts identified via visual
inspection, EEG data were further inspected for artifacts with independent component analysis (ICA) and dipole analysis (Delorme and
Makeig 2004; Hammon et al. 2008). Extended Infomax ICA allowed
for identification of temporally maximally independent component
(IC) time courses that could be categorized into muscular and oculomotor activity as well as cortically based processes (e.g., see Gwin
et al. 2010). ICs were analyzed with respect to scalp topography and
frequency characteristics during the epoched time intervals, and those
that displayed features indicative of artifacts were excluded from the
back-projection to sensor space. Ocular ICs were identified according
to the criteria described in the literature (Delorme and Makeig 2004;
Jung et al. 1998): smoothly decreasing EEG spectra; a strong farfrontal projection; and large voltage fluctuations corresponding to
individual eye movements. ICs representing muscle artifacts were
identified as having a spatially focal scalp projection and high power
at high frequencies (20 –50 Hz and above). Dipole models that best
explained the scalp topography were fit to each of the remaining
components with the DIPFIT plug-in for EEGLAB and localized
within a four-shell spherical boundary element head model (BEM) of
the Montreal Neurological Institute standard brain (an average of 152
MRI brain scans of healthy individuals). We morphed the digitized 3D
electrode array to the head model by scaling and rotating the head
coordinate system so that the digitized anatomical reference points
were aligned with the head model anatomical reference points as
provided by the DIPFIT 2 plug-in for EEGLAB (see http://sccn.ucsd.
edu/wiki/A08:_DIPFIT) and as used elsewhere (Gwin et al. 2010,
2011; Gwin and Ferris 2012). All ICs but one were fitted with a single
dipole, resulting in minimal residual variances between IC scalp map
and dipole scalp projection, following the procedure described previously (Delorme et al. 2012; Gwin et al. 2010, 2011; Makeig et al.
2002; Sipp et al. 2013). Across all subjects, the residual variance of
one of the occipital ICs could be further reduced by a dual dipole fit
(with a symmetrical constraint right/left), since its component topography indicated a bilateral distribution. However, for the clustering
(see below), EEGLAB automatically selected only the right or left
dipole aspect, preventing the multiple use of a dual dipole for
clustering. Only ICs whose dipoles resided within the vicinity of the
EEG CORRELATES OF SPATIAL COGNITIVE MAPPING
across condition, cluster, and latency (Hothorn et al. 2008). Significance was defined as P ⬍ 0.05 for all parametric tests.
Source estimation. Standardized low-resolution brain electromagnetic
tomography (sLORETA, http://www.uzh.ch/keyinst/loreta.htm) was
used to estimate the cortical 3D distribution of current density based
on the scalp topography (Fuchs et al. 2002; Pascual-Marqui 2002;
Pascual-Marqui et al. 1994). For ERPs, sLORETA was computed
using the mean amplitude during a 40-ms interval centered on the
peak of the ERP difference wave between misplaced and correctly
placed objects. Additionally, the following procedure complemented
the surface-based reconstruction of distributed sources: After IC
decomposition, all single-trial IC time series (⫺2,025 to 1,975 ms
time-locked to bubble pop) of the identified brain components as well
as components capturing oculomotor processes were transformed into
a spectrographic image using Morlet wavelets in the frequency range
between 2 and 55 Hz (EEGLAB functions std_precomp and newtimef
using the following parameters and values: “freqs” ⫽ [2 55], “timelimits” ⫽ [⫺2,025 1,975], “baseline” ⫽ 0, “cycles” ⫽ [3 0.5],
“padratio” ⫽ 1, and “ntimesout” ⫽ 400). Each single-trial component
activity time series was then transformed to a spectrographic image,
yielding for each trial the total power changes (see David et al. 2006
for details). The event-related spectral perturbations (ERSPs; Klimesch et al. 2007) of each trial were baseline corrected and averaged.
To identify sets of similar components across all subjects, IC
processes were clustered with a k-means cluster algorithm (Hartigan
and Wong 1979) implemented in the EEGLAB STUDY toolbox. For
this purpose, an N-dimensional cluster position vector for each IC was
created to measure “distances” between all ICs in the defined cluster
space based on particular IC characteristics such as IC log spectra,
single-trial ERPs, IC inverse weights across the electrodes, equivalent
dipole locations, single-trial ERSPs, and single-trial intertrial coherence (ITC), or phase locking factor (Onton et al. 2006). Parameters
were chosen to emphasize 3D location of sources and spectral perturbation/coherence that we expected to be most relevant given the
present task: A weighting factor of 1 was assigned to the IC log
spectrum (2–55 Hz) and to the inverse weight topography, a weighting
factor of 3 was assigned to the single-trial ERP waveforms of the IC,
a weighting factor of 4 was assigned to the single-trial ERSP and ITC
data, and a weighting factor of 15 was assigned to the dipole location.
This set of weighting factors has been shown in the literature to
balance the relative influence of the various IC measures, so that no
one measure has disproportionate influence on cluster assignment (see
Gramann et al. 2010; Plank et al. 2010). The resulting joint vector was
reduced to 10 principal dimensions with principal component analysis
(PCA) (Jung et al. 2001) and was set to 7 final clusters based on
existing studies on spatial cognition using the same ICA and clustering approach (Gramann et al. 2010; Plank et al. 2010). Of the seven
defined clusters, two clusters emerged that reflected ocular processes
that were common to all subjects (horizontal eye movements and
blinks); the remaining five clusters could be localized in the cortical
domain. The two ocular clusters were retained and fed into the same
processing pipeline as the cortical clusters in order to determine
whether differences between experimental conditions would be specifically related to cortical ICs but not eye movements. Components
that were located further than 3 standard deviations from any of the
cluster centroids were classified as outliers and removed from further
analysis. For sLORETA reconstructions of the resulting IC cluster
centroids, the mean topographical weights of all ICs contributing to a
cluster were used. The cluster-based findings are presented in ERSP
Results in Source Space.
In contrast to standard EEGLAB functionality, where the computation of ERSP differences simply takes the actual difference between
conditions and shuffles the pixels (the number of shuffles is determined by the significance threshold), we masked the actual differences between misplaced and correctly placed objects with the randomization-based mask according to Maris and Oostenveld (2007)
described above, revealing differences statistically different from a
J Neurophysiol • doi:10.1152/jn.00114.2014 • www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.5 on June 18, 2017
eyeballs (ocular ICs) or within the brain volume of the head model
and with no more than 15% residual variance between IC scalp map
and dipole scalp projection were retained. On average, 9.7 ⫾ 2.8
(mean ⫾ SD) ocular and brain-based ICs per participant were identified (range 4 –14). Cleaned EEG data were generated by backprojecting the time course of activity within the remaining ICs to the
surface electrodes. The ocular ICs were excluded from the backprojection in order to remove artifacts due to eye movements. However, the ocular ICs were retained for the analysis in source space;
since the clustering procedure can isolate ICs related to eye movements consistently across subjects into specific clusters, this permitted
us to determine whether differences between experimental conditions
are related to eye movements (see below). This procedure allows
attenuation of artifacts from the EEG without having to reject the
entire trial during which an artifact occurred (Jung et al. 2000).
ERP difference waves. Artifact-free component activations were
back-projected to the electrodes and baseline corrected (⫺225 to ⫺25
ms). Average ERP waveforms were computed for correctly placed
versus misplaced objects, and differences waves were generated. The
amplitude difference waves of misplaced minus correctly placed
objects across the 64 cap electrodes were submitted to a randomization-based masking procedure referred to as the cluster mass permutation test (Bullmore et al. 1999; Groppe et al. 2011; Maris and
Oostenveld 2007; Worsley et al. 1996). This nonparametric testing
procedure was selected because of its robustness against outliers and
different trial numbers for correctly placed and misplaced objects. In
this procedure, the t-score of the actual grand average difference
waveform as generated by averaging the individual difference waves
between misplaced and correctly placed objects is compared with a
distribution of t-scores of difference waves of surrogate grand averages without knowing their assignment to misplaced and correctly
placed conditions. This procedure was accomplished as described in
Groppe et al. (2011). Differences in the real grand average difference
amplitudes were considered significant for t-scores corresponding to a
global ␣-level of 5%.
The large number of comparisons in the point-by-point bootstrapping procedure introduces false positives across both electrodes (see
below) and time. We assume that temporal false positives occur as a
Poisson-like process, without any pattern, but the already complicated
signal is temporally continuous (from earlier filtering) and we cannot
expect parametric statistics. Continuity means that “neighboring”
significant points are more likely to be significant just because the data
have been filtered to remove large differences, even if the significance
is spurious. The goal is to estimate the minimum duration that
separates “neighbors” (neighbor time). Thus we bootstrap the duration
of spurious intervals (10,000 resamples from the difference amplitudes above) by swapping the “correctly placed” and “misplaced”
labels within each electrode and finding the longest contiguous time
interval with apparent significance. The upper 95% confidence bound
of these durations estimates the neighbor time (␣ ⫽ 0.05). The
average neighbor time was 26 ⫾ 4 ms (mean ⫾ SD), which corresponds to frequency of 39(6) Hz, or somewhat slower than the
high-pass filter at 55 Hz, as we would expect. Any significant intervals
in the original masked data with duration shorter than the neighbor
time were removed. The approach for temporal bootstrapping described above only applies to multiple comparisons within an electrode, but false positives may still exist across electrodes. To simplify
the problem across electrodes we clustered them into 10 regions (see
ERP Results in Sensor Space below) and used parametric tests on the
peak values at latencies of interest via a linear mixed model in R
(version 3.0.1) and the lme4 package version 1.0-6 (Bates et al. 2014).
The clustering was necessary for the linear mixed models to converge.
To test for significant difference, models with and without fixed
effects were compared in marginal order with a ␹2-test on the
difference in profiled deviance (log likelihood penalized for the
number of terms in the model). Post hoc tests were accomplished by
using the multcomp package to control for multiple comparisons
743
744
EEG CORRELATES OF SPATIAL COGNITIVE MAPPING
distribution of mean surrogate differences (for each time-frequency
bin, or “pixel”) derived from randomized experimental conditions.
Whenever the actual ERSP difference exceeded 1.96 standard deviations of the randomized ERSP distribution, we classified the ERSP
difference between misplaced and correctly placed objects as statistically significant (P ⬍ 0.05).
lack of explicit learning instructions on day 1. Furthermore, a
subject’s rating of how interesting the individual objects were
did not predict subsequent memory of the object/location
relations, and there were no marked differences in interest
across objects.
ERP Results in Sensor Space
RESULTS
Given that on day 2 subjects had to select between one of
two alternative choices (correctly placed vs. misplaced),
chance performance was 50%. On day 2 subjects performed
well above chance, correctly identifying 86.86% ⫾ 5.94%
(mean ⫾ SD) of all objects as misplaced or correctly placed
compared with day 1 (minimum ⫽ 76.60%, maximum ⫽
95.90%). No relationship between interest ratings on day 1 and
performance on day 2 was found (r ⫽ ⫺0.0035, P ⫽ 0.99;
Spearman test). Additionally, an ANOVA was performed to
measure dependence of the rated interest of the objects across
subjects and showed that no object was consistently rated as
either interesting or not [P ⫽ 0.2, bootstrap KolmogorovSmirnov (bKS) test detailed in Snider et al. 2013b]. Similarly,
no location was more memorable than any other (P ⫽ 0.85,
bKS test). In sum, subjects performed well above chance,
despite a delay of 24 h between exposure and testing and the
7
9
6
10
8
**
9
6
3
10
7
8
**
3
4
**
correctly placed
misplaced
-200
0
200
400
600
800
voltage (µV)
**
-3
-2
-1
0
1
2
3
4
**
0
*
*
2
**
**
2
4
-200
5
5
1
1
*
200
400
600
latency (ms)
800
-200
0
200
400
600
800
Fig. 2. Grand average ERP waveforms evoked by correctly placed vs. misplaced objects in specified electrode clusters in the time interval ⫺225 to ⫹975 ms
time-locked to bubble pop: 1 and 2, left and right occipito-parietal; 3, 4, and 5, left, medial, and right centro-parietal; 6, 7, and 8, left, medial, and right
fronto-central; 9 and 10, left and right frontal. While the N160 effect (light gray area) is significant in posterior clusters 1, 2, and 5, the N535 effect (dark gray
area) is present in all electrode clusters 1– 6 and 8 (all except anterior regions). Significant differences: *P ⬍ 0.05; **P ⬍ 0.01.
J Neurophysiol • doi:10.1152/jn.00114.2014 • www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.5 on June 18, 2017
Figure 2 presents the grand average ERP waveforms evoked
by correctly placed versus misplaced objects in specified electrode clusters: 1: left occipito-parietal (PO3, PO7, O1), 2: right
occipito-parietal (PO4, PO8, O2), 3: left centro-parietal (CP3,
CP5, TP7, P3, P5, P7), 4: medial centro-parietal (CPz, Pz), 5:
right centro-parietal (CP4, CP6, TP8, P4, P6, P8), 6: left
fronto-central (FC1, FC3, FC5, C1, C3, C5), 7: medial frontocentral (FCz, Cz), 8: right fronto-central (FC2, FC4, FC6, C2,
C4, C6), 9: left frontal (F1, F3, F5), and 10: right frontal (F2,
F4, F6).
To initially identify differences in the ERP waveform to
correctly placed or misplaced objects, we identified two peaks
of activity in the grand average difference wave centered at
around 160 and 535 ms after the object was revealed. We ran
a linear mixed model with random effects allowing for subjectlevel variability in response to condition (correctly placed or
misplaced), electrode cluster, and peak latency (N160 or
N535). Consistent with the waveforms shown in Fig. 2, the
Behavioral Results
EEG CORRELATES OF SPATIAL COGNITIVE MAPPING
745
Table 1. Mean amplitudes for N160 and N535 ERP components at each of the 10 electrode clusters
No.
Electrode Cluster
N160
Misplaced, ␮V
N160 Correctly Placed, ␮V
N535
Misplaced, ␮V
N535 Correctly Placed, ␮V
1
2
3
4
5
6
7
8
9
10
L occipito-parietal
R occipito-parietal
L centro-parietal
M centro-parietal
R centro-parietal
L fronto-central
M fronto-central
R fronto-central
L frontal
R frontal
⫺0.66 ⫾ 0.59**
⫺0.92 ⫾ 0.61**
⫺0.22 ⫾ 0.34
⫺0.10 ⫾ 0.70*
⫺0.29 ⫾ 0.33
0.49 ⫾ 0.38
0.92 ⫾ 0.76
0.63 ⫾ 0.36
0.56 ⫾ 0.59
0.74 ⫾ 0.57
⫺1.30 ⫾ 0.49**
⫺1.68 ⫾ 0.51**
⫺0.51 ⫾ 0.27
⫺0.56 ⫾ 0.55*
⫺0.79 ⫾ 0.27
0.40 ⫾ 0.34
0.73 ⫾ 0.68
0.37 ⫾ 0.32
0.59 ⫾ 0.57
0.62 ⫾ 0.54
⫺0.41 ⫾ 0.40**
⫺0.15 ⫾ 0.43**
⫺0.90 ⫾ 0.25**
⫺0.69 ⫾ 0.54**
⫺0.45 ⫾ 0.25*
⫺1.17 ⫾ 0.28**
⫺1.00 ⫾ 0.56
⫺0.86 ⫾ 0.27*
⫺1.36 ⫾ 0.41
⫺1.07 ⫾ 0.42
0.23 ⫾ 0.28**
0.45 ⫾ 0.31**
⫺0.27 ⫾ 0.17**
0.06 ⫾ 0.38**
0.07 ⫾ 0.18*
⫺0.55 ⫾ 0.19**
⫺0.38 ⫾ 0.38
⫺0.33 ⫾ 0.18*
⫺0.80 ⫾ 0.28
⫺0.58 ⫾ 0.28
Values are mean ⫾ SD amplitudes for the N160 and N535 event-related potential (ERP) components at each of the 10 electrode clusters. Significant differences
between correctly placed and misplaced objects within each electrode cluster: *P ⬍ 0.05, **P ⬍ 0.01.
µV
- 0.7
N515-555
0.8
0
µV
midline
0
In addition to changes in ERPs, we hypothesized that theta
would differ between conditions. To test this hypothesis, we
examined time-frequency dynamics by means of ERSPs at the
level of clustered ICs. As mentioned in MATERIALS AND METHODS, on
average, 9.7 ⫾ 2.8 (mean ⫾ SD) ocular and brain-based ICs
per participant were identified (range: 4 –14) and used for
clustering, resulting in the specifications of the cluster centroids shown in Table 2.
Cluster 1 comprised ICs whose inverse weights projected
most strongly toward frontal midline electrodes and exhibited
power in the (3– 6 Hz) low theta frequency range, displaying
well-established characteristics of “frontal midline theta”
(Hsieh and Ranganath 2014) (Fig. 5, A and C). By sLORETA,
the scalp distributions of these components were consistent
with generation in medial frontal cortex (BA 6) and/or the
anterior cingulate gyrus (BA 32, 24; Talairach coordinates X,
N515-555
1
0.5
0 µV
-200
- 0.8
N145-175
left hemisphere
0.7
ERSP Results in Source Space
right hemisphere
N145-175
Increased amplitude of N535 over left hemisphere and
central parietal regions for misplaced objects. In addition,
there were significant ERP amplitude differences in the time
range 515–555 ms apparent in a large number of electrodes (49
of 64 electrodes), peaking in medial parietal scalp leads and
expanding toward bilateral central and left frontal electrodes.
This difference was due to an increased negativity for misplaced objects in the given time interval. Estimated sources of
the difference wave by sLORETA were most prominent in
bilateral fusiform gyrus (BA 19, 37) and/or lingual gyrus (BA
18; Talairach coordinates X, Y, Z ⫽ ⫺19, ⫺60, ⫺8; Fig. 4B).
-0.5
-1
0
200
400
600
800
latency [ms]
J Neurophysiol • doi:10.1152/jn.00114.2014 • www.jn.org
Fig. 3. Cluster mass permutation statistics of
the difference wave between grand averages
of misplaced and correctly placed objects,
time-locked to bubble pop (time 0). EEG
channels from anterior to posterior (left and
right hemispheres and midline) are plotted
along the y-axis. Topographical maps on the
left represent the mean difference amplitude
between misplaced and correctly placed objects in the time windows between 145 and
175 ms (top) and between 515 and 555 ms
(bottom). Significant electrodes (P ⬍ 0.05)
are highlighted with white circles.
Downloaded from http://jn.physiology.org/ by 10.220.33.5 on June 18, 2017
results of the linear mixed model indicated that the triple
interaction of condition, cluster, and peak latency was highly
significant [␹2(9) ⫽ 28.36, P ⬍ 0.001]. Given the structure of
the experiment, we interpret this interaction as a difference in
activity between conditions (correctly placed or misplaced
object) with variable strength depending on electrode cluster
and peak latency. Post hoc tests (as indicated in Fig. 2) bore
this out, with the earlier N160 component showing an
⬃0.5-␮V increased negativity over lateral parieto-occipital
regions for correctly placed objects. The N535 showed widespread differences extending from occipital to fronto-central
regions with ⬃0.5-␮V increased negativity for misplaced objects (opposite in polarity compared with the N160 effect).
Table 1 provides the mean amplitudes for the N160 and N535
components at each of the 10 electrode clusters.
Decrease in centro-parietal N160 amplitude for misplaced
objects. The mean amplitude differences between conditions
were examined in more detail with nonparametric tests over the
entire space of electrodes and times. Permutation statistics
(Fig. 3) revealed a significant difference in ERP amplitudes
between correctly placed and misplaced conditions in the time
interval 145–175 ms after bubble pop. This effect of a decreased negativity for misplaced objects was strongest in electrodes over central and bilateral occipital/parietal regions.
Source estimates of the difference wave as returned by
sLORETA were consistent with generation in medial parietal
areas, particularly precuneus (BA 7/31; Talairach coordinates
X, Y, Z ⫽ ⫺9, ⫺52, 29; Fig. 4A).
746
EEG CORRELATES OF SPATIAL COGNITIVE MAPPING
A
B
0.0210
0 µV
0.0155
- 0.7
0
0.8
0.182
0 µV
0.091
- 0.8
0
Y, Z ⫽ ⫺4, 11, 37) (Fig. 5B). Based on the permutation-based
masking procedure described in MATERIALS AND METHODS,
group-level ERSPs, that is, event-locked changes in spectral
power for this source, could be computed in component space
for correctly placed and misplaced objects (Fig. 5D). Figure 5D
shows that there is increased delta and theta power shortly after
the bubble pop for both expected and misplaced objects. This
increase in total power was significantly greater for misplaced
objects within the low theta frequency range (3– 6 Hz) between
145 and 475 ms after bubble pop. No other cluster exhibited
differences between conditions in the theta range, including
cluster 2, which localized to BA 18/31 (very close to the
previously observed N160). Cluster 2 also showed an ERP
difference with increased negativity to correctly placed objects
peaking at 170 ms, similar to the N160 component.
DISCUSSION
Basic Findings
The present study investigated neural processes underlying
learning of the spatial location of objects when subjects freely
walked about a complex, naturalistic, large-scale virtual environment. Because of technical advances, we were able to
record event-related EEG in conjunction with behavioral measures during naturalistic, full-body exploration in 3D space.
We experimentally probed acquired spatial representations by
altering the location of a subset of objects from those previously learned in an unsupervised fashion. Our primary hypothesis was that amplitude modulations during the N400 time
interval would differentiate brain responses to objects placed in
the expected location versus in an unexpected location. Al-
Table 2. Specifications of IC clusters identified by the k-means clustering procedure
Talairach Coordinates
Cluster
VEOG
HEOG
1
2
3
4
5
X
⫺4
0
20
⫺30
⫺10
Y
11
⫺72
⫺22
20
⫺12
Z
37
22
43
13
56
Brodmann Area
BA
BA
BA
BA
BA
32/24
18/31
24/31
13/45
6/31
Vertical eye movement cluster
Horizontal eye cluster
Left anterior cingulate gyrus
Bilateral (pre)cuneus
Right cingulate gyrus
Left inferior frontal gyrus
Left paracentral lobule/medial frontal gyrus
Ss
ICs
12
12
12
11
8
9
10
14
14
24
17
16
17
14
Values are specifications of the independent component (IC) clusters identified by the k-means clustering procedure, including X, Y, Z locations of the cluster
centroid specified within the stereotaxic coordinate system of Talairach and Tournoux (1988) and their anatomical region defined in the Brodmann area system
(Brodmann 1925). Nos. of subjects (Ss) and ICs within each cluster are also provided.
J Neurophysiol • doi:10.1152/jn.00114.2014 • www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.5 on June 18, 2017
Fig. 4. Standardized low-resolution brain electromagnetic tomography (sLORETA) source
estimates of the difference wave between the
grand averages of misplaced and correctly
placed objects in the time window 145–175
ms peaking in the precuneus (BA 7/31; Talairach coordinates X, Y, Z ⫽ ⫺9, ⫺52, 29)
(A) and 515–555 ms peaking in the bilateral
fusiform gyrus (BA 19, 37) and/or lingual
gyrus (BA 18; Talairach coordinates X, Y, Z ⫽
⫺19, ⫺60, ⫺8) (B). Significant electrodes are
highlighted with white circles.
0.7
EEG CORRELATES OF SPATIAL COGNITIVE MAPPING
A
+
B
0.094
0
0.047
-
0
frequency [Hz]
CORRECTLY PLACED
MISPLACED
DIFFERENCE(abs(z) > 1.96)
36
1.5
22
1
0.5
0 dB
-0.5
-1
-1.5
14
8
5
3
-1000 -500
0
500 1000
latency [ms]
-1000 -500
0
500 1000
4
Fig. 5. A: average topography generated by
mapping all inverse weights of independent
components (ICs) most strongly weighted in
frontal midline electrodes and exhibiting
spectral power in the (3– 6 Hz) low theta
frequency range. B: sLORETA source estimates localize in the medial frontal cortex
(BA 6) and/or anterior cingulate (BA 32, 24;
Talairach coordinates X, Y, Z ⫽ ⫺4, 11, 37).
C: results of the equivalent dipole fitting procedure as implemented in EEGLAB. Individual IC sources are colored blue; the cluster
centroid is colored red. D: event-related spectral perturbations (ERSPs) of correctly placed
(left) and misplaced (center) objects as well as
permutation-masked, z-transformed differences between the 2 conditions (right). Time
is plotted on the x-axis (in ms relative to
bubble pop); frequencies are plotted along the
y-axis. Colors indicate power (dB), with
warmer colors representing an increase in
power and colder colors indicating a decrease
in power.
2
0
-2
-4
-1000 -500
latency [ms]
though this ERP component has mainly been studied with
words (Kutas and Federmeier 2011), it has also been observed
to other meaningful complex stimuli such as faces (Eimer
2000), mathematical computations (Galfano et al. 2004),
sounds (Aramaki et al. 2010; Painter and Koelsch 2011), and
visual scenes (Amoruso et al. 2013; Ganis and Kutas 2003). As
N400 amplitude has been conceptualized to reflect cognitive
processes coordinating the “feed-forward flow of stimulusdriven activity with a state of the distributed, dynamically
active neural landscape that is (semantic) memory” (Kutas and
Federmeier 2011), it is seemingly modulated by the contextual
congruity of a stimulus with its surroundings. Expanding these
conceptualizations to the spatial domain, we predicted an
N400-like component to be elicited when objects have to be
integrated with their spatial context and its amplitude to be
increased as participants’ spatial expectancies of object-location associations were violated and objects were encountered
whose locations were interchanged from those learned on the
previous day. This hypothesis was supported: We found an
N400-like component that resembled lexico-semantic N400s in
morphology and scalp topography, with some differences in
peak latency. Source estimates were consistent with an increased negativity being generated in ventral posterior temporal areas.
Previous studies in rodents and humans have shown the
central role of theta oscillation for active maintenance and
retrieval of mental representations (Buzsaki 2005; Cornwell et
0
500 1000
latency [ms]
al. 2008; Klimesch 1999; O’Keefe and Burgess 1999; Sauseng
et al. 2002), particularly in tasks involving spatial attention and
memory (Bischof and Boulanger 2003; Caplan et al. 2003; de
Araujo et al. 2002; Jones and Wilson 2005; Kahana et al.
1999). In humans, frontal-midline theta power is modulated by
task demands (Jensen and Tesche 2002), and we expected an
increase in theta power when encountering alterations to a
previously learned environment. This hypothesis was supported: misplaced objects elicited an increase in frontal-midline theta power, particularly in the 3–5 Hz range. Source
estimates of temporally maximally independent components
based on single dipole fitting in a BEM as well as distributed
source estimation in sLORETA were consistent with this
increase originating in anterior cingulate cortex.
Finally, we found a negative-going deflection occurring
⬃160 ms after presentation of an object that differentiated
between correctly placed and misplaced objects, resembling an
N200, which has been linked to cognitive processes monitoring
stimulus frequency (Sahin et al. 2009; Squires et al. 1975) and
broad semantic category (Chan et al. 2011; Pulvermuller et al.
1999). In contrast to semantic studies in which the N200 was
found to originate from inferior frontal or temporal regions,
source estimates in our paradigm suggested generators in
medial parietal cortices.
By analogy with possibly homologous components observed
during verbal processing, the early N200-like component may
reflect first-pass processing of object-location information
J Neurophysiol • doi:10.1152/jn.00114.2014 • www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.5 on June 18, 2017
C
D
747
748
EEG CORRELATES OF SPATIAL COGNITIVE MAPPING
along the dorsal stream. The later N400-like component would
reflect the cognitive comparison of the actual object location
with a memory-based representation of its expected location in
the environment. Finally, the late theta activity may organize
retrieval of spatial information from the mental representation.
An N400-Like ERP Component Reflects Contextual
Integration in the Spatial Domain
Differences Between Conditions Are Observable as Early as
145 ms After Stimulus
N400 amplitude is influenced by a wide range of distant
associations, across sensory modalities, semantic domains, and
memory systems (Brown and Hagoort 1993; Halgren 1990;
Holcomb et al. 2002; Kutas and Federmeier 2011). Consistent
with this broad access to widely distributed information, laminar recordings have found that the N400 is generated by
J Neurophysiol • doi:10.1152/jn.00114.2014 • www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.5 on June 18, 2017
Our results indicate the presence of an N400-like ERP
component peaking in midline parietal contacts and extending
toward temporal regions that is elicited by violation of objectlocation expectations that had been learned in an unsupervised
fashion on the previous day. Its morphology, temporal characteristics, and spatial distribution suggest similarities with the
N400 for semantic stimuli (Kutas and Federmeier 2011). Neural sources of the N400, as identified by intracranial recordings
and MEG, have been localized in temporal areas (e.g., anterior
medial, middle superior, and ventral temporal) and dorsolateral
prefrontal cortex, bilaterally, but primarily in the left hemisphere (Dale et al. 2000; Halgren et al. 1994a, 1994b, 2002;
Nobre et al. 1994; Smith et al. 1986; Tse et al. 2007; Van
Petten and Luka 2006). These findings have been complemented by hemodynamic studies suggesting involvement of
left superior temporal and left inferior frontal gyri (e.g., Van
Petten and Luka 2006).
The present finding that the reconstructed source of our
N400-like component was ventral temporal regions may derive
from the cognitive demands for learning and recalling spatial
locations and associated objects from an internal representation. As noted by O’Keefe and Nadel (1978), the spatial
cognitive map is highly abstract and flexible, since the world
around us is constantly changing, objects in the environment
might vanish, or viewing conditions might change from one
encounter to the next (Christou and Bülthoff 2000). We note
that subjects in the present task learned novel objects and their
locations not simply from sequences of stimuli presented on a
two-dimensional screen but rather from active, unsupervised
spatial exploration in an immersive virtual reality environment.
Subjects approached the objects from different angles and were
required to view them from different perspectives in order to
accomplish the task. The identification of object location correspondences approached from different viewpoints and cognitive demands for recalling spatial locations required associative processes in order to bind object features such as shape,
color, contrast, etc. and extract multidimensional object features such as identity and location. The present source estimates of the N400-like ERP component suggest the involvement of extrastriate areas that play a major role for object
identity (Kohler et al. 1998; Moscovitch et al. 1995) and object
recognition (Price et al. 1996; Rosier et al. 1997; Stewart et al.
2001; Zelkowicz et al. 1998), in concert with the hippocampal
formation further down the ventral stream. Interestingly, several studies have pointed out the critical role of Brodmann area
37 (occipital-temporal cortex) for the detection of violations of
spatial expectancies (e.g., Faillenot et al. 1997; Goel et al.
2004; Smith et al. 1995).
Another aspect to be discussed is the difference in latency
between the classic N400 component and the N400-like component identified in the present study at 515–555 ms. Kutas and
Federmeier (2011) characterize the N400 as a monophasic
negativity between 200 and 600 ms, peaking at 400 ms after
stimulus onset (relative to a 100 ms prestimulus baseline). In
semantic paradigms, the N400 amplitude was shown to begin
at around 250 ms, peak at around 400 ms, and return to
baseline at around 600 ms after stimulus (e.g., Kutas and
Hillyard 1980). In the study of Eimer (2000), negativities were
apparent between 300 and 500 ms after face onset. In studies
using mathematical computations (Galfano et al. 2004), the
core N400 time window was defined between 300 and 400 ms,
whereas in paradigms utilizing complex visual scenes (Amoruso et al. 2013; Ganis and Kutas 2003; Sitnikova et al. 2003,
2008), N400-like waves were found to be present in the time
range between 325 and 600 ms. One of the major reasons for
latency shifts and modulations in the duration of the effect may
be stimulus complexity. In studies that utilize rapid stimulus
sequences with brief presentations, the N400 seems to occur
earlier than in studies using complex stimuli. In the present
paradigm, continuous visual input, self-determined choice of
perspective on the task-related objects, and subject-driven
duration of exploration might have been accompanied by a
comparable shift in component latency due to the prolonged
presentation of the objects, allowing for extended interaction
that unfolded over several hundred milliseconds (see Sitnikova
et al. 2008).
A second issue relates to the duration of the classic N400
versus that found in the present study. As mentioned above,
depending upon the type of stimuli presented, the duration of
the N400 varied from about 200 ms to 400 ms. We found
significant differences between conditions of correctly placed
and misplaced objects in the time interval 515–555 ms; however, the difference was nonzero for ⬃200 ms (Fig. 2). This
duration is in the same range as that found for other complex
stimuli.
Taken together, the functional sensitivity, latency range,
topographic distribution, as well as the estimated source locations suggest a classification of the present findings in the
functional family of the N400. We agree with the notion put
forward by Kutas and Federmeier (2011, p. 623) that the label
“N400-like” is solely used as a “heuristic label for stimulusrelated brain activity in the 200 – 600 ms poststimulus-onset
window with a characteristic morphology and, critically, a
pattern of sensitivity to experimental variables—and hence a
common functionality,” which also applies in the context of
our paradigm. Since comparable N400 effects have been identified for nonlinguistic stimuli such as faces (Halgren et al.
1994a, 1994b), mathematical computations (Galfano et al.
2004), static visual scenes (Amoruso et al. 2013; Ganis and
Kutas 2003; McPherson and Holcomb 1999), and complex
video stimuli (Sitnikova et al. 2003, 2008), the present results
support an interpretation of the N400 as reflecting a general
cognitive process across different semantic domains.
EEG CORRELATES OF SPATIAL COGNITIVE MAPPING
Frontal Midline Theta Is Sensitive to Object-Location
Mismatches
Our paradigm might be considered as a form of a spatial
paired-associate learning task in which participants learned
associations between objects and locations. Paired-associate
learning has been demonstrated not only when subjects are
presented with a stimulus item and must reproduce the associated pair item (Yim et al. 2013) but also when only recognition of the pair item is required (Bunsey and Eichenbaum
1996; Ragland et al. 1995). EEG studies of paired-associate
learning have generally identified modulations of the contingent negative variation (CNV) and P300 ERP components over
central-parietal areas (Peters et al. 1977). In the present study,
theta power was modulated by the experimental manipulation.
Although there are clear similarities of our paradigm to pairedassociate learning, there also are differences. In our paradigm,
learning was accomplished in an unsupervised fashion, i.e.,
spatial information was acquired incidentally, without explicit
allocation of attentional resources toward learning object-location associations. During the learning session in day 1 of the
experiment, subjects did not know that they would later be
asked to recall object-location associations. In contrast, in
paired-associate learning, learning is accomplished explicitly
with foreseeable consequences for successful or unsuccessful
retrieval (Schacter et al. 1998).
We observed total spectral power between 3 and 6 Hz in
electrodes over midline frontal regions to be modulated by
misplacing objects. Specifically, misplaced objects elicited
increased synchronizations of high delta/low theta during 145–
475 ms after stimulus onset. This finding was based on an
across-subject statistical clustering of individual ICs based on
specific component properties such as ERPs, ERSPs, ITCs, and
estimated dipole location in a BEM. Unlike other IC clusters
that localized in different brain regions and showed effects in
other frequency ranges, this cluster alone exhibited significant
power in the theta range and dissociated correctly placed and
misplaced objects. Equivalent dipole models of the ICs as well
as distributed source estimation in sLORETA were compatible
with areas in the anterior cingulate sulcus and/or the overlying
prefrontal cortex, consistent with other studies utilizing comparable component-based clustering procedures (Gramann et
al. 2010; Michels et al. 2008; Onton et al. 2005; Plank et al.
2010). This localization is distinct from the evoked response
that occurred at a similar time (N535) but was located anteriorly in the bilateral fusiform gyrus and/or lingual gyrus. With
other experimental paradigms, theta has been found to emanate
not only from medial and lateral prefrontal but also from
central, parietal, and medial temporal cortices (Cornwell et al.
2008; Hsieh and Ranganath 2014; Kahana et al. 1999; Raghavachari et al. 2001; Wang et al. 2005). EEG studies in humans
have estimated (6 – 8 Hz) theta during memory-related tasks to
originate in or near medial prefrontal cortex, particularly in
anterior cingulate cortex (Asada et al. 1999; Gevins and Smith
2000; Meltzer et al. 2008), but it has been shown that during
spatial processing lower theta and delta effects (3– 6 Hz) are
also observable (Watrous et al. 2011). This was apparent in
Snider et al. (2013b), where spatial autocorrelation during
walking was found to be strongest in the 2– 8 Hz (delta/theta)
range. Moreover, these findings are consistent with human
intracranial studies of the hippocampus, where low-frequency
(2– 6 Hz) modulations also have been observed (Arnolds et al.
1979; Brazier 1968; Jacobs et al. 2007; Mormann et al. 2008;
Rutishauser et al. 2010; Watrous et al. 2011).
Theta in general has been associated with a wide range of
cognitive processes and has been found to correlate with
memory retrieval, locomotion, and goal-directed behavior (Caplan et al. 2003; Cornwell et al. 2008; Ekstrom et al. 2003;
Kahana et al. 2001; Klimesch 1999). Particularly in situations
with high cognitive demands and working memory load increased levels of theta power have been identified (Bastiaansen
et al. 2002; Klimesch 1999; Meltzer et al. 2008; Onton et al.
2005; Raghavachari et al. 2001; Sauseng et al. 2002; Schack et
al. 2005; Tesche and Karhu 2000), which can be interpreted in
concordance with findings on event-related synchronization
(ERS) effects in the theta range (Michels et al. 2008; Pfurtscheller and Aranibar 1977, 1979). In the present study we
found increased frontal midline theta power for incorrectly
placed objects. This would be consistent with misplaced objects eliciting increased cognitive demands for memory recall
and detecting deviations between currently encountered and
internally represented environmental configurations in frontal
regions. This finding extends the existing literature on increased frontal midline theta power associated with increased
cognitive demands.
Furthermore, the crucial role of theta for spatial learning and
memory in humans is well documented (Bischof and Boulanger 2003; Caplan et al. 2003; de Araujo et al. 2002; Jones
and Wilson 2005; Kahana et al. 1999). In an intracranial study
of Kahana and colleagues (1999), increased theta was detected
during recall compared with learning, as well as when participants were confronted with more complex environments.
Bischof and Boulanger (2003) also found increased theta in the
time range when subjects encountered central navigation points
along a previously learned route that were crucial for successful task accomplishment. Theta oscillations are seemingly
modulated by the amount of self-motion cues available to the
subject. While attenuated in stationary setups or under bodily
restraints, theta power increases as subjects are being passively
moved along trajectories or experience visual motion cues in
the absence of proprioceptive or vestibular signals, with the
J Neurophysiol • doi:10.1152/jn.00114.2014 • www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.5 on June 18, 2017
activity in upper cortical layers reflecting second-pass associative processing (Halgren et al. 2006). The same sites that
generate the N400 often generate a negativity at ⬃200 ms in
layer 4 that reflects first-pass feedforward processing (Halgren
et al. 2006). This component can be influenced by word
characteristics such as lexical frequency (Hauk et al. 2006;
Sahin et al. 2009) and even general semantic category (Chan et
al. 2011). In the present study, misplaced objects evoked ERP
differences in a component peaking at ⬃160 ms, suggesting
that the expected spatial location of an object may be a
fundamental characteristic processed in the first sweep of
activity through the cortex. This rapid processing of object
location would have clear adaptive advantage in responding
rapidly within a changing spatial environment. Medial parietal
and retrosplenial areas were estimated as the possible source
for this effect. This region, strongly related to both medial
temporal and parietal cortices, may play a key role in relating
allocentric and egocentric spaces, to guide adaptive interaction
with the outside world (Cho and Sharp 2001; Maguire 2001;
Vann and Aggleton 2004).
749
750
EEG CORRELATES OF SPATIAL COGNITIVE MAPPING
more individualized IC selection procedures with stronger
physiological validity.
We cannot rule out an evoked component to the spectral
differences we observed for IC cluster 1 (theta cluster). However, we note three findings in favor of it being induced. First,
the theta cluster did not match any evoked component observed
in the electrode clusters. Second, frontal midline theta has a
long-standing role in the literature (as described above) that is
consistent with the role in memory proposed here. Finally, the
theta cluster localization is far removed from the localization of
the evoked activity at the same time.
Conclusions
Taken together, the present results suggest that the increased
negativity for object-location mismatches in the time range
between 400 and 700 ms peaking over parietal regions that
localizes in temporal regions resembles an N400 in semantic
memory paradigms, here reflecting the mismatch between
sensory input and representational expectation. Second, the
increase in frontal midline theta power in case of violations of
object-location expectancies might be related to recalling the
actual object as learned on the previous day. Both effects (theta
power and N400 amplitude) are most likely generated by an
extensive network comprising posterior parietal and anterior
cingulate cortex, potentially mediated by hippocampus.
ACKNOWLEDGMENTS
We thank Jamie Lukos and Jason Trees for help with the data collection.
Present address of Markus Plank: Brain Products GmbH, Zeppelinstr. 7,
82205 Gilching, Germany.
GRANTS
Supported in part by Office of Naval Research MURI Award No. N0001410-1-0072 and National Science Foundation Grants SMA-1041755 and ENG1137279 (EFRI M3C).
Limitations
The analysis pipeline used here was a major component
enabling the collection of usable EEG during ambulatory
movement, but it does have some limitations. One potential
limitation is the assumption that an IC can be localized to a
dipolar source within the brain. Indeed, a split occipital IC was
observed in the data presented here that could only be fit with
two dipoles (and it was). We then removed any other ICs
whose best-fit dipole was outside the brain. This processing
pipeline followed procedures that have been promulgated by
other authors (Delorme et al. 2012; Gwin et al. 2010, 2011;
Makeig et al. 2002; Sipp et al. 2013), but this could lead to loss
of brain signal (Castellanos and Makarov 2006).
While ICA decomposition has been shown to be a highly
useful tool for EEG data analysis and noise attenuation, assumptions of ICA and dipole source modeling might not map
1:1 into physiological processes, as discussed by Delorme and
colleagues (2012). For example, the assumption that scalp
potentials are generated by synchronization across localized
cortical patches might represent a simplification of spatiotemporal nonstationary processes, phase-coherent activations
across cortical regions or hemispheres, which cannot be handled well with dipole-based reconstruction methods. More
accurate head models, for example, based on segmentation of
actual structural MRIs of the participants, or frequency-domain
complex ICA decomposition techniques might allow for even
DISCLOSURES
No conflicts of interest, financial or otherwise, are declared by the author(s).
AUTHOR CONTRIBUTIONS
Author contributions: M.P., J.S., E.K., E.H., and H.P. conception and
design of research; M.P. and J.S. performed experiments; M.P., J.S., E.K.,
E.H., and H.P. analyzed data; M.P., J.S., E.K., E.H., and H.P. interpreted
results of experiments; M.P. and J.S. prepared figures; M.P., J.S., E.H., and
H.P. drafted manuscript; M.P., J.S., E.K., E.H., and H.P. edited and revised
manuscript; M.P., J.S., E.K., E.H., and H.P. approved final version of manuscript.
REFERENCES
American Electroencephalographic Society. Guidelines for standard electrode position nomenclature. J Clin Neurophysiol 8: 200 –202, 1991.
Amoruso L, Gelormini C, Aboitiz F, Alvarez Gonzalez M, Manes F,
Cardona JF, Ibanez A. N400 ERPs for actions: building meaning in
context. Front Hum Neurosci 7: 57, 2013.
Aramaki M, Marie C, Kronland-Martinet R, Ystad S, Besson M. Sound
categorization and conceptual priming for nonlinguistic and linguistic
sounds. J Cogn Neurosci 22: 2555–2569, 2010.
Arnolds DE, Lopes da Silva FH, Aitink JW, Kamp A. Hippocampal EEG
and behaviour in dog. III. Hippocampal EEG correlates of stimulus-response
tasks and of sexual behaviour. Electroencephalogr Clin Neurophysiol 46:
581–591, 1979.
J Neurophysiol • doi:10.1152/jn.00114.2014 • www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.5 on June 18, 2017
strongest theta response being generated as subjects are allowed to move freely, generating consistent visual, vestibular,
and proprioceptive input (Russell et al. 2006; Terrazas et al.
2005; Vanderwolf 1975).
We believe that our experimental paradigm allowed subjects
to fully utilize multisensory information, allowing for naturalistic integration and consolidation of the acquired knowledge
of the environment into a robust internal representation based
on unsupervised spatial learning. Since subjects had to recall
spatial information on object location traversing the same
environment again on day 2, theta activity might establish a
dynamic framework for distributed cognitive processing between spatially segregated cortical areas (Mizuhara et al.
2004), allowing for action-oriented retrieval of spatial information from the internal representation. In nonspatial experiments, theta localized to the anterior cingulate has been associated with cognitive conflict (Kovacevic et al. 2012). Our
observation of an increase in theta to misplaced objects may
thus reflect sustained retrieval and reprocessing of spatial
information resulting from conflicts between incoming sensory
input and representational systems. The modulations observable in frontal midline theta are certainly distinct from the
N400 effects, since they differ in latency and localization of the
source estimates. However, future research will have to address
how representational recall as represented by frontal midline
theta relates to hippocampal-parietal processing of objectlocation information as reflected by the N400 effects. Given
the large literature on sex differences in spatial orientation
(Coluccia and Louse 2004; Lawton and Morrin 1999), future
studies should also examine potential sex differences using
naturalistic data acquisition environments such as that in the
present experiment.
EEG CORRELATES OF SPATIAL COGNITIVE MAPPING
Galfano G, Mazza V, Angrilli A, Umilta C. Electrophysiological correlates
of stimulus-driven multiplication facts retrieval. Neuropsychologia 42:
1370 –1382, 2004.
Ganis G, Kutas M. An electrophysiological study of scene effects on object
identification. Brain Res Cogn Brain Res 16: 123–144, 2003.
Gevins A, Smith ME. Neurophysiological measures of working memory and
individual differences in cognitive ability and cognitive style. Cereb Cortex
10: 829 – 839, 2000.
Goel V, Makale M, Grafman J. The hippocampal system mediates logical
reasoning about familiar spatial environments. J Cogn Neurosci 16: 654 –
664, 2004.
Gramann K, Onton J, Riccobon D, Müller HJ, Bardins S, Makeig S.
Human brain dynamics accompanying the use of egocentric and allocentric
reference frames during spatial navigation. J Cogn Neurosci 22: 2836 –2849,
2010.
Groppe DM, Urbach TP, Kutas M. Mass univariate analysis of event-related
brain potentials/fields. I. A critical tutorial review. Psychophysiology 48:
1711–1725, 2011.
Gwin JT, Ferris DP. An EEG-based study of discrete isometric and isotonic
human lower limb muscle contractions. J Neuroeng Rehabil 9: 9 –35, 2012.
Gwin JT, Gramann K, Makeig S, Ferris DP. Removal of movement artifact
from high-density EEG recorded during walking and running. J Neurophysiol 103: 3526 –3534, 2010.
Gwin JT, Gramann K, Makeig S, Ferris DP. Electrocortical activity is
coupled to gait cycle phase during treadmill walking. Neuroimage 54:
1289 –1296, 2011.
Hafting T, Fyhn M, Molden S, Moser MB, Moser EI. Microstructure of a
spatial map in the entorhinal cortex. Nature 436: 801– 806, 2005.
Halgren E. Insights from evoked potentials into the neuropsychological
mechanisms of reading. In: Neurobiology of Higher Cognitive Function,
edited by Scheibeland AB, Wechsler AF. New York: Guilford, 1990, p.
103–150.
Halgren E, Baudena P, Heit G, Clarke JM, Marinkovic K. Spatio-temporal
stages in face and word processing. 1. Depth-recorded potentials in the
human occipital, temporal and parietal lobes. J Physiol (Paris) 88: 1–50,
1994a.
Halgren E, Baudena P, Heit G, Clarke JM, Marinkovic K, Chauvel P.
Spatio-temporal stages in face and word processing. 2. Depth-recorded
potentials in the human frontal and Rolandic cortices. J Physiol (Paris) 88:
51– 80, 1994b.
Halgren E, Dhond RP, Christensen N, Van Petten C, Marinkovic K,
Lewine JD, Dale AM. N400-like magnetoencephalography responses modulated by semantic context, word frequency, and lexical class in sentences.
Neuroimage 17: 1101–1116, 2002.
Halgren E, Wang C, Schomer DL, Knake S, Marinkovic K, Wu J, Ulbert
I. Processing stages underlying word recognition in the anteroventral temporal lobe. Neuroimage 30: 1401–1413, 2006.
Hammon PS, Makeig S, Poizner H, Todorov E, de Sa VR. Predicting
reaching targets from human EEG. Signal Processing Mag IEEE 25: 69 –77,
2008.
Hartigan JA, Wong MA. Algorithm AS 136: a k-means clustering algorithm.
Appl Stat 28: 100 –108, 1979.
Hasselmo ME, Stern CE. Theta rhythm and the encoding and retrieval of
space and time. Neuroimage 852: 656 – 666, 2014.
Hauk O, Davis MH, Ford M, Pulvermuller F, Marslen-Wilson WD. The
time course of visual word recognition as revealed by linear regression
analysis of ERP data. Neuroimage 30: 1383–1400, 2006.
Heit G, Smith ME, Halgren E. Neural encoding of individual words and
faces by the human hippocampus and amygdala. Nature 333: 773–775,
1988.
Holcomb PJ, Grainger J, O’Rourke T. An electrophysiological study of the
effects of orthographic neighborhood size on printed word perception. J
Cogn Neurosci 14: 938 –950, 2002.
Hothorn T, Bretz F, Westfall P. Simultaneous inference in general parametric models. Biom J 50: 346 –363, 2008.
Hsieh LT, Ranganath C. Frontal midline theta oscillations during working
memory maintenance and episodic encoding and retrieval. Neuroimage 85:
721–729, 2014.
Jacobs J, Kahana MJ, Ekstrom AD, Fried I. Brain oscillations control
timing of single-neuron activity in humans. J Neurosci 27: 3839 –3844,
2007.
Jensen O, Tesche CD. Frontal theta activity in humans increases with memory
load in a working memory task. Eur J Neurosci 15: 1395–1399, 2002.
J Neurophysiol • doi:10.1152/jn.00114.2014 • www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.5 on June 18, 2017
Asada H, Fukuda Y, Tsunoda S, Yamaguchi M, Tonoike M. Frontal
midline theta rhythms reflect alternative activation of prefrontal cortex and
anterior cingulate cortex in humans. Neurosci Lett 274: 29 –32, 1999.
Bastiaansen MC, Posthuma D, Groot PF, de Geus EJ. Event-related alpha
and theta responses in a visuo-spatial working memory task. Clin Neurophysiol 113: 1882–1893, 2002.
Bates D, Maechler M, Bolker B, Walker S. lme4: Linear mixed-effects models
using Eigen and S4. R package version 1.0 – 6. http://CRAN.R-project.org/
packageⴝlme4. (08/03/2014).
Bischof WF, Boulanger P. Spatial navigation in virtual reality environments:
an EEG analysis. Cyberpsychol Behav 6: 487– 495, 2003.
Brazier MA. Studies of the EEG activity of limbic structures in man.
Electroencephalogr Clin Neurophysiol 25: 309 –318, 1968.
Brodmann K. Vergleichende Lokalisationslehre der Grosshirnrinde: In ihren
Principien dargestellt auf Grund des Zellenbaues. Leipzig, Germany: Barth,
1925.
Brown C, Hagoort P. The processing nature of the N400: evidence from
masked priming. J Cogn Neurosci 5: 34 – 44, 1993.
Bullmore ET, Suckling J, Overmeyer S, Rabe-Hesketh S, Taylor E,
Brammer MJ. Global, voxel, and cluster tests, by theory and permutation,
for a difference between two groups of structural MR images of the brain.
IEEE Trans Med Imaging 18: 32– 42, 1999.
Bunsey M, Eichenbaum H. Conservation of hippocampal memory function in
rats and humans. Nature 379: 255–257, 1996.
Buzsaki G. Theta rhythm of navigation: link between path integration and
landmark navigation, episodic and semantic memory. Hippocampus 15:
827– 840, 2005.
Caplan JB, Madsen JR, Schulze-Bonhage A, Aschenbrenner-Scheibe R,
Newman EL, Kahana MJ. Human theta oscillations related to sensorimotor integration and spatial learning. J Neurosci 23: 4726 – 4736, 2003.
Castellanos NP, Makarov VA. Recovering EEG brain signals: artifact suppression with wavelet enhanced independent component analysis. J Neurosci Methods 158: 300 –312, 2006.
Chan AM, Baker JM, Eskandar E, Schomer D, Ulbert I, Marinkovic K,
Cash SS, Halgren E. First-pass selectivity for semantic categories in human
anteroventral temporal lobe. J Neurosci 31: 18119 –18129, 2011.
Chance SS, Gaunet F, Beall A, Loomis JM. Locomotion mode affects the
updating of objects encountered during travel: the contribution of vestibular
and proprioceptive inputs to path integration. Presence 7: 168 –178, 1998.
Cho JW, Sharp PE. Head direction, place, and movement correlates for cells
in the rat retrosplenial cortex. Behav Neurosci 115: 3–25, 2001.
Christou C, Bülthoff HH. Perception, representation and recognition: a
holistic view of recognition. Spat Vis 13: 265–275, 2000.
Coluccia E, Louse G. Gender differences in spatial orientation: a review. J
Environ Psychol 24: 329 –340, 2004.
Cornwell BR, Johnson LL, Holroyd T, Carver FW, Grillon C. Human
hippocampal and parahippocampal theta during goal-directed spatial navigation predicts performance on a virtual Morris water maze. J Neurosci 28:
5983–5990, 2008.
Dale AM, Liu AK, Fischl BR, Buckner RL, Belliveau JW, Lewine JD,
Halgren E. Dynamic statistical parametric mapping: combining fMRI and
MEG for high-resolution imaging of cortical activity. Neuron 26: 55– 67,
2000.
David O, Kilner JM, Friston KJ. Mechanisms of evoked and induced
responses in MEG/EEG. Neuroimage 31: 1580 –1591, 2006.
de Araujo DB, Baffa O, Wakai RT. Theta oscillations and human navigation:
a magnetoencephalography study. J Cogn Neurosci 14: 70 –78, 2002.
Delorme A, Makeig S. EEGLAB: an open source toolbox for analysis of
single-trial EEG dynamics including independent component analysis. J
Neurosci Methods 134: 9 –21, 2004.
Delorme A, Palmer J, Onton J, Oostenveld R, Makeig S. Independent EEG
sources are dipolar. PLoS One 7: e30135, 2012.
Eimer M. Event-related brain potentials distinguish processing stages involved in face perception and recognition. Clin Neurophysiol 111: 694 –705,
2000.
Ekstrom AD, Kahana MJ, Caplan JB, Fields TA, Isham EA, Newman EL,
Fried I. Cellular networks underlying human spatial navigation. Nature
425: 184 –188, 2003.
Faillenot I, Toni I, Decety J, Gregoire MC, Jeannerod M. Visual pathways
for object-oriented action and object recognition: functional anatomy with
PET. Cereb Cortex 7: 77– 85, 1997.
Fuchs M, Kastner J, Wagner M, Hawes S, Ebersole JS. A standardized
boundary element method volume conductor model. Clin Neurophysiol 113:
702–712, 2002.
751
752
EEG CORRELATES OF SPATIAL COGNITIVE MAPPING
Muller RU, Kubie JL, Ranck JB Jr. Spatial firing patterns of hippocampal
complex-spike cells in a fixed environment. J Neurosci 7: 1935–1950, 1987.
Nobre AC, Allison T, McCarthy G. Word recognition in the human inferior
temporal lobe. Nature 372: 260 –263, 1994.
O’Keefe J, Burgess N. Theta activity, virtual navigation and the human
hippocampus. Trends Cogn Sci 3: 403– 406, 1999.
O’Keefe J, Burgess N, Donnett JG, Jeffery KJ, Maguire EA. Place cells,
navigational accuracy, and the human hippocampus. Philos Trans R Soc
Lond B Biol Sci 353: 1333–1340, 1998.
O’Keefe J, Nadel L. The Hippocampus as a Cognitive Map. Oxford, UK:
Oxford Univ. Press, 1978.
Oldfield RC. Ambidexterity in surgeons. Lancet 1: 655, 1971.
Onton J, Delorme A, Makeig S. Frontal midline EEG dynamics during
working memory. Neuroimage 27: 341–356, 2005.
Onton J, Westerfield M, Townsend J, Makeig S. Imaging human EEG
dynamics using independent component analysis. Neurosci Biobehav Rev
30: 808 – 822, 2006.
Painter JG, Koelsch S. Can out-of-context musical sounds convey meaning?
An ERP study on the processing of meaning in music. Psychophysiology 48:
645– 655, 2011.
Pascual-Marqui RD. Standardized low-resolution brain electromagnetic tomography (sLORETA): technical details. Methods Find Exp Clin Pharmacol 24, Suppl. D: 5–12, 2002.
Pascual-Marqui RD, Michel CM, Lehmann D. Low resolution electromagnetic tomography: a new method for localizing electrical activity in the
brain. Int J Psychophysiol 18: 49 – 65, 1994.
Patterson K, Nestor PJ, Rogers TT. Where do you know what you know?
The representation of semantic knowledge in the human brain. Nat Rev
Neurosci 8: 976 –987, 2007.
Peters JF, Billinger TW, Knott JR. Event related potentials of brain (CNV
and P300) in a paired associate learning paradigm. Psychophysiology 14:
579 –585, 1977.
Pfurtscheller G, Aranibar A. Event-related cortical desynchronization detected by power measurements of scalp EEG. Electroencephalogr Clin
Neurophysiol 42: 817– 826, 1977.
Pfurtscheller G, Aranibar A. Evaluation of event-related desynchronization
(ERD) preceding and following voluntary self-paced movement. Electroencephalogr Clin Neurophysiol 46: 138 –146, 1979.
Plank M, Müller HJ, Onton J, Makeig S, Gramann K. Human EEG
correlates of egocentric and allocentric path integration. Spat Cogn 6222:
191–206, 2010.
Price CJ, Moore CJ, Humphreys GW, Frackowiak RS, Friston KJ. The
neural regions sustaining object recognition and naming. Proc Biol Sci 263:
1501–1507, 1996.
Pulvermuller F, Lutzenberger W, Preissl H. Nouns and verbs in the intact
brain: evidence from event-related potentials and high-frequency cortical
responses. Cereb Cortex 9: 497–506, 1999.
Raghavachari S, Kahana MJ, Rizzuto DS, Caplan JB, Kirschen MP,
Bourgeois B, Madsen JR, Lisman JE. Gating of human theta oscillations
by a working memory task. J Neurosci 21: 3175–3183, 2001.
Ragland JD, Gur RC, Deutsch GK, Censits DM, Gur RE. Reliability and
construct validity of the Paired-Associate Recognition Test: a test of declarative memory using Wisconsin Card Sorting stimuli. Psychol Assess 7:
25–32, 1995.
Rosier A, Cornette L, Dupont P, Bormans G, Michiels J, Mortelmans L,
Orban GA. Positron-emission tomography imaging of long-term shape
recognition challenges. Proc Natl Acad Sci USA 94: 7627–7632, 1997.
Russell NA, Horii A, Smith PF, Darlington CL, Bilkey DK. Lesions of the
vestibular system disrupt hippocampal theta rhythm in the rat. J Neurophysiol 96: 4 –14, 2006.
Rutishauser U, Ross IB, Mamelak AN, Schuman EM. Human memory
strength is predicted by theta-frequency phase-locking of single neurons.
Nature 464: 903–907, 2010.
Sahin NT, Pinker S, Cash SS, Schomer D, Halgren E. Sequential processing
of lexical, grammatical, and phonological information within Broca’s area.
Science 326: 445– 449, 2009.
Sauseng P, Klimesch W, Gruber W, Doppelmayr M, Stadler W, Schabus
M. The interplay between theta and alpha oscillations in the human electroencephalogram reflects the transfer of information between memory
systems. Neurosci Lett 324: 121–124, 2002.
Schack B, Klimesch W, Sauseng P. Phase synchronization between theta and
upper alpha oscillations in a working memory task. Int J Psychophysiol 57:
105–114, 2005.
J Neurophysiol • doi:10.1152/jn.00114.2014 • www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.5 on June 18, 2017
Jones MW, Wilson MA. Theta rhythms coordinate hippocampal-prefrontal
interactions in a spatial memory task. PLoS Biol 3: 2187–2199, 2005.
Jung TP, Humphries C, Lee TW, Makeig S, McKeown MJ, Iragui V,
Sejnowski TJ. Extended ICA removes artifacts from electroencephalographic recordings. Adv Neur In 10: 894 –900, 1998.
Jung TP, Makeig S, Humphries C, Lee TW, McKeown MJ, Iragui V,
Sejnowski TJ. Removing electroencephalographic artifacts by blind source
separation. Psychophysiology 37: 163–178, 2000.
Jung TP, Makeig S, Westerfield M, Townsend J, Courchesne E, Sejnowski
TJ. Analysis and visualization of single-trial event-related potentials. Hum
Brain Mapp 14: 166 –185, 2001.
Kahana MJ, Seelig D, Madsen JR. Theta returns. Curr Opin Neurobiol 11:
739 –744, 2001.
Kahana MJ, Sekuler R, Caplan JB, Kirschen M, Madsen JR. Human theta
oscillations exhibit task dependence during virtual maze navigation. Nature
399: 781–784, 1999.
Klatzky RL. Allocentric and egocentric spatial representations: definitions,
distinctions, and interconnections. In: Lecture Notes in Artificial Intelligence, Spatial Cognition I, edited by Freksa C, Habel C, Wender KF.
Heidelberg, Germany: Springer, 1998, p. 1–17.
Klimesch W. EEG alpha and theta oscillations reflect cognitive and memory
performance: a review and analysis. Brain Res Rev 29: 169 –195, 1999.
Klimesch W, Sauseng P, Hanslmayr S. EEG alpha oscillations: the inhibition-timing hypothesis. Brain Res Rev 53: 63– 88, 2007.
Kohler S, Moscovitch M, Winocur G, Houle S, McIntosh AR. Networks of
domain-specific and general regions involved in episodic memory for spatial
location and object identity. Neuropsychologia 36: 129 –142, 1998.
Kovacevic S, Azma S, Irimia A, Sherfey J, Halgren E, Marinkovic K.
Theta oscillations are sensitive to both early and late conflict processing
stages: effects of alcohol intoxication. PLoS One 7: e43957, 2012.
Kuperberg GR. Neural mechanisms of language comprehension: challenges
to syntax. Brain Res 1146: 23– 49, 2007.
Kutas M, Federmeier KD. Thirty years and counting: finding meaning in the
N400 component of the event-related brain potential (ERP). Annu Rev
Psychol 62: 621– 647, 2011.
Kutas M, Hillyard SA. Reading senseless sentences: brain potentials reflect
semantic incongruity. Science 207: 203–205, 1980.
Lawton CA, Morrin KA. Gender differences in pointing accuracy in computer-simulated 3D mazes. Sex Roles 40: 73–92, 1999.
Maguire EA. The retrosplenial contribution to human navigation: a review of
lesion and neuroimaging findings. Scand J Psychol 42: 225–238, 2001.
Makeig S, Westerfield M, Jung TP, Enghoff S, Townsend J, Courchesne
E, Sejnowski TJ. Dynamic brain sources of visual evoked responses.
Science 295: 690 – 694, 2002.
Marinkovic K, Baldwin S, Courtney MG, Witzel T, Dale AM, Halgren E.
Right hemisphere has the last laugh: neural dynamics of joke appreciation.
Cogn Affect Behav Neurosci 11: 113–130, 2011.
Marinkovic K, Rosen BQ, Cox B, Kovacevic S. Event-related theta power
during lexical-semantic retrieval and decision conflict is modulated by
alcohol intoxication: anatomically constrained MEG. Front Psychol 3: 121,
2012.
Maris E, Oostenveld R. Nonparametric statistical testing of EEG- and
MEG-data. J Neurosci Methods 164: 177–190, 2007.
McPherson WB, Holcomb PJ. An electrophysiological investigation of
semantic priming with pictures of real objects. Psychophysiology 36: 53– 65,
1999.
Meltzer J, Zaveri HP, Goncharova I, Distasio MM, Papademetris X,
Spencer SS, Spencer DD, Constable RT. Effects of working memory load
on oscillatory power in human intracranial EEG. Cereb Cortex 18: 1843–
1855, 2008.
Michels L, Moazami-Goudarzi M, Jeanmonod D, Sarnthein J. EEG alpha
distinguishes between cuneal and precuneal activation in working memory.
Neuroimage 40: 1296 –1310, 2008.
Mizuhara H, Wang LQ, Kobayashi K, Yamaguchi Y. A long-range cortical
network emerging with theta oscillation in a mental task. Neuroreport 15:
1233, 2004.
Mormann F, Osterhage H, Andrzejak RG, Weber B, Fernandez G, Fell J,
Elger CE, Lehnertz K. Independent delta/theta rhythms in the human
hippocampus and entorhinal cortex. Front Hum Neurosci 2: 3, 2008.
Moscovitch C, Kapur S, Kohler S, Houle S. Distinct neural correlates of
visual long-term memory for spatial location and object identity: a positron
emission tomography study in humans. Proc Natl Acad Sci USA 92:
3721–3725, 1995.
EEG CORRELATES OF SPATIAL COGNITIVE MAPPING
Talairach J, Tournoux P. Co-planar Stereotaxic Atlas of the Human Brain.
New York: Thieme Medical, 1988.
Terrazas A, Krause M, Lipa P, Gothard KM, Barnes CA, McNaughton
BL. Self-motion and the hippocampal spatial metric. J Neurosci 25: 8085–
8096, 2005.
Tesche CD, Karhu J. Theta oscillations index human hippocampal activation
during a working memory task. Proc Natl Acad Sci USA 97: 919 –924, 2000.
Tse CY, Lee CL, Sullivan J, Garnsey SM, Dell GS, Fabiani M, Gratton G.
Imaging cortical dynamics of language processing with the event-related
optical signal. Proc Natl Acad Sci USA 104: 17157–17162, 2007.
Van Petten C, Luka BJ. Neural localization of semantic context effects in
electromagnetic and hemodynamic studies. Brain Lang 97: 279 –293, 2006.
Vanderwolf CH. Neocortical and hippocampal activation relation to behavior:
effects of atropine, eserine, phenothiazines, and amphetamine. J Comp
Physiol Psychol 88: 300 –323, 1975.
Vann SD, Aggleton JP. Testing the importance of the retrosplenial guidance
system: effects of different sized retrosplenial cortex lesions on heading
direction and spatial working memory. Behav Brain Res 155: 97–108, 2004.
Wang C, Ulbert I, Schomer DL, Marinkovic K, Halgren E. Responses of
human anterior cingulate cortex microdomains to error detection, conflict
monitoring, stimulus-response mapping, familiarity, and orienting. J Neurosci 25: 604 – 613, 2005.
Watrous AJ, Fried I, Ekstrom AD. Behavioral correlates of human hippocampal delta and theta oscillations during navigation. J Neurophysiol 105:
1747–1755, 2011.
Worsley KJ, Marrett S, Neelin P, Vandal AC, Friston KJ, Evans AC. A
unified statistical approach for determining significant signals in images of
cerebral activation. Hum Brain Mapp 4: 58 –73, 1996.
Yim H, Dennis SJ, Sloutsky VM. The development of episodic memory:
items, contexts, and relations. Psychol Sci 24: 2163–2172, 2013.
Zelkowicz BJ, Herbster AN, Nebes RD, Mintun MA, Becker JT. An
examination of regional cerebral blood flow during object naming tasks. J
Int Neuropsychol Soc 4: 160 –166, 1998.
J Neurophysiol • doi:10.1152/jn.00114.2014 • www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.5 on June 18, 2017
Schacter DL, Buckner RL, Koutstaal W. Memory, consciousness and
neuroimaging. Philos Trans R Soc Lond B Biol Sci 353: 1861–1878, 1998.
Siapas AG, Lubenov EV, Wilson MA. Prefrontal phase locking to hippocampal theta oscillations. Neuron 46: 141–151, 2005.
Sipp AR, Gwin JT, Makeig S, Ferris DP. Loss of balance during balance
beam walking elicits a multifocal theta band electrocortical response. J
Neurophysiol 110: 2050 –2060, 2013.
Sitnikova T, Holcomb PJ, Kiyonaga KA, Kuperberg GR. Two neurocognitive mechanisms of semantic integration during the comprehension of
visual real-world events. J Cogn Neurosci 20: 2037–2057, 2008.
Sitnikova T, Kuperberg G, Holcomb PJ. Semantic integration in videos of
real-world events: an electrophysiological investigation. Psychophysiology
40: 160 –164, 2003.
Smith DM, Mizumori SJ. Hippocampal place cells, context, and episodic
memory. Hippocampus 16: 716 –729, 2006.
Smith EE, Jonides J, Koeppe RA, Awh E, Schumacher EH, Minoshima S.
Spatial versus object working memory: PET investigations. J Cogn Neurosci
7: 337–356, 1995.
Smith ME, Stapleton JM, Halgren E. Human medial temporal lobe potentials evoked in memory and language tasks. Electroencephalogr Clin Neurophysiol 63: 145–159, 1986.
Snider J, Plank M, Lee D, Poizner H. Simultaneous neural and movement
recording in large-scale immersive virtual environments. IEEE Trans
Biomed Circuits Syst 7: 713–721, 2013a.
Snider J, Plank M, Lynch G, Halgren E, Poizner H. Human cortical theta
during free exploration encodes space and predicts subsequent memory. J
Neurosci 33: 15056 –15068, 2013b.
Solstad T, Boccara CN, Kropff E, Moser MB, Moser EI. Representation of
geometric borders in the entorhinal cortex. Science 322: 1865–1868, 2008.
Squires NK, Squires KC, Hillyard SA. Two varieties of long-latency positive
waves evoked by unpredictable auditory stimuli in man. Electroencephalogr
Clin Neurophysiol 38: 387– 401, 1975.
Stewart L, Meyer B, Frith U, Rothwell J. Left posterior BA37 is involved in
object recognition: a TMS study. Neuropsychologia 39: 1– 6, 2001.
753