Cerebral Cortex August 2014;24:2229–2236
doi:10.1093/cercor/bht084
Advance Access publication March 31, 2013
Working Memory Coding of Analog Stimulus Properties in the Human Prefrontal Cortex
Bernhard Spitzer1, Matthias Gloel2, Timo T. Schmidt2,3 and Felix Blankenburg1,3
1
Dahlem Institute for Neuroimaging of Emotion, Freie Universität Berlin, Berlin 14195, Germany, 2Bernstein Center for
Computational Neuroscience, Berlin 10115, Germany and 3Center for Adaptive Rationality (ARC), Max Planck Institute for
Human Development, Berlin 14195, Germany
Address Correspondence to B. Spitzer, Dahlem Institute for Neuroimaging of Emotion, Freie Universität Berlin, Habelschwerdter Allee 45, 14195
Berlin, Germany. Email: [email protected]
Building on evidence for working memory (WM) coding of vibrotactile frequency information in monkey prefrontal cortex, recent
electroencephalography studies found frequency processing in
human WM to be reflected by quantitative modulations of prefrontal
upper beta activity (20–30 Hz) as a function of the to-be-maintained
stimulus attribute. This kind of stimulus-dependent activity has
been observed across different sensory modalities, suggesting a
generalized role of prefrontal beta during abstract WM processing
of quantitative magnitude information. However, until now the available empirical evidence for such quantitative WM representation
remains critically limited to the retention of periodic stimulus frequencies. In the present experiment, we used retrospective cueing
to examine the quantitative WM processing of stationary (intensity)
and temporal (duration) attributes of a previously presented tactile
stimulus. We found parametric modulations of prefrontal beta
activity during cued WM processing of each type of quantitative
information, in a very similar manner as had before been observed
only for periodic frequency information. In particular, delayed prefrontal beta modulations systematically reflected the magnitude of
the retrospectively selected stimulus attribute and were functionally
linked to successful behavioral task performance. Together, these
findings converge on a generalized role of stimulus-dependent prefrontal beta-band oscillations during abstract scaling of analog
quantity information in human WM.
Keywords: EEG, oscillations, somatosensory, stimulus coding, working
memory
Introduction
It is widely accepted that prefrontal cortex (PFC) plays a key
role in working memory (WM), that is, in operations enabling
the maintenance and online processing of information that no
longer exists in the environment (for reviews, see Miller and
D’Esposito 2005; Pasternak and Greenlee 2005; D’Esposito
2007). Studies of human WM, mostly in the visual and verbal
domains, routinely reported PFC activity in terms of hemodynamic responses (Wager and Smith 2003; Nee et al. 2012), but
also in terms of changes in oscillatory activity as assessed by
electro- or magnetoencephalography (EEG/MEG) (e.g., TallonBaudry et al. 1998; for review, see Benchenane et al. 2011).
Thereby, WM function was typically examined by contrasting
conditions of high versus no (or lower) WM processing, yielding insights into task-dependent activity associated with different WM operations and materials (Gazzaley and Nobre 2012).
Another line of research in the somatosensory domain has
recently succeeded in delineating direct, stimulus-dependent
signatures of WM processing in the human PFC. In particular,
complementing evidence for parametric WM coding of vibrotactile frequency information in the monkey PFC (Romo et al.
© The Author 2013. Published by Oxford University Press. All rights reserved.
For Permissions, please e-mail: [email protected]
1999; Barak et al. 2010), human EEG studies found WM processing of different frequencies to be systematically reflected
by parametric modulations of right prefrontal oscillatory
activity in the upper beta band (20–30 Hz; Spitzer et al. 2010;
Spitzer and Blankenburg 2011). Recently, this prefrontal
activity has also been observed in other sensory modalities
(Spitzer and Blankenburg 2012), suggesting a general role of
parametric beta oscillations during WM processing of analog
stimulus quantities. However, the currently available empirical evidence for such parametric WM processing remains
restricted to tasks that require the maintenance of periodic
frequency information, thus leaving open whether modulations of prefrontal beta may represent other stimulus
attributes, as well. Moreover, the precise nature of the information, which is parametrically encoded during WM processing of different frequencies, is not clear yet. Periodic stimuli
of different temporal frequency, as defined by the number of
cyclic repeats per second, differ in their fine-grained temporal
properties and might be evaluated in terms of the perceived
speed of the stimulus dynamics. However, different stimulus
frequencies may also be evaluated with respect to the inferred
intensity of stimulation, with faster oscillating stimuli
suggesting the presence of higher physical energy. It therefore remains unknown to what extent frequency-dependent
parametric modulations of prefrontal beta activity might
reflect a unique concept of periodic stimulus frequency, or
the processing of temporal structure or physical energy, or a
more general abstract representation of sensory quantity.
To shed light on these questions, in the present experiment,
we systematically examined the WM processing of temporal
(duration) and stationary (intensity) quantitative attributes of
to-be-maintained vibrotactile stimuli. We used a modified
delayed-match-to-sample (DMTS) task (Fig. 1A), which involved
retrospective selection of the task-relevant stimulus attribute for
sustained maintenance, allowing us to disambiguate the active
WM processing of either stimulus dimension after identical encoding conditions. Examining prefrontal EEG oscillations, we
hypothesized that if parametric WM effects (Romo et al. 1999;
Spitzer et al. 2010; Spitzer and Blankenburg 2012) were uniquely linked to a proprietary concept of periodic frequencies,
no parametric modulations, neither by duration nor by intensity, should be observed. If, on the other hand, frequencydependent modulations relate to the processing of temporal
stimulus properties, then prefrontal beta activity might be
modulated—if at all—by stimulus duration, but not by intensity.
In contrast, if the previous frequency-dependent modulations
intrinsically reflect a scaling of perceived physical energy, independent of its precise temporal structure, then a modulation by
intensity only is expected. If however, as proposed recently
(Spitzer and Blankenburg 2012), parametric WM processing in
with the presentation of a base stimulus (s1). The amplitude
(vibration intensity) of s1 was randomly chosen to be 39.1%, 53.7%,
68.4%, or 83.0% of the stimulator’s maximum capacity (corresponding
to 1600, 2200, 2800, and 3400 arbitrary amplitude units of 4095 maximally available units). The resulting stimulation intensities were not
painful, and ranged between the 26- and 56-fold of the average detection threshold for this type of stimuli (1.5% stimulator capacity, or
60.3 amplitude units; assessed in a follow-up test on 5 of the participants using the method-of-limits procedure). The duration of s1 was
randomly chosen to be 350, 450, 550, or 650 ms, independent of s1’s
intensity. Eight hundred milliseconds after offset of s1, a visual cue
(“A” or “B,” randomly selected on each trial) indicated whether subjects should for the remainder of the trial focus on s1’s intensity
(A-cue) or duration (B-cue). After a 2500-ms retention interval, a
comparison stimulus (s2) was presented, which differed from s1 both
in intensity (by ±7.3% stimulator capacity, corresponding to ±300 amplitude units) and in duration (by ±150 ms). In the A-cue condition,
subjects’ task was to indicate whether s2 was weaker or stronger than
s1 by pressing a response button with the right hand either once or
twice. Analogously, in the B-cue condition, subjects were asked to
indicate whether s2 was shorter or longer than s1. After several practice trials, each participant completed 7 blocks of 64 trials, for a total
of 224 discrimination trials in each cue condition.
Figure 1. (A) Experimental task. On each trial, a vibrotactile stimulus (s1) of a given
intensity and duration was presented at the left index finger. A visual cue instructed
subjects to retrospectively focus on s1’s intensity (“A”) or duration (“B”) throughout a
retention interval (green), for subsequent comparison against the corresponding
attribute of s2. (B) Upper left: Time frequency statistical parametric maps of
differences in oscillatory activity between the A- and B-tasks, collapsed across all
values of s1. Dashed gray rectangles indicate TF window and channels where a
significant effect was observed (P < 0.05, FWE). Lower left: Time-course of beta
(15–25 Hz) activity changes in the A- and B-tasks, for the channels outlined above.
Right: SPM 3D source reconstruction of the difference in beta power between 700
and 1300 ms.
the PFC relies on a generalized WM representation of quantity,
then prefrontal upper beta may be modulated by either of the 2
stimulus attributes, when their abstract magnitude is actively
processed in WM.
EEG Recording and Analysis
EEG was recorded using a 64-channel active electrode system (ActiveTwo, BioSemi, Amsterdam, Netherlands), with electrodes placed in
an elastic cap according to the extended 10-20 system. Individual electrode locations were registered using a stereotactic electrode positioning system (Zebris Medical GmbH, Isny, Germany). Vertical and
horizontal eye movements were recorded from 4 additional channels.
Signals were digitized at a sampling rate of 2048 Hz, off-line bandpass filtered (1–100 Hz), downsampled to 512 Hz, and average referenced. All analyses were carried out using SPM8 for MEG/EEG
(update revision number 4667, 27 February 2012; Wellcome Department of Cognitive Neurology, London, UK: www.fil.ion.ucl.ac.uk/
spm/) and custom MATLAB code (The Mathworks, Inc., Natick, USA).
The EEG was corrected for eye blinks using calibration data to generate individual artifact coefficients and adaptive spatial filtering (for
details, see Ille et al. 2002). Remaining artifacts were rejected by
excluding all data containing signal amplitudes >80 µV from analysis.
Materials and Methods
Subjects
Twenty-six healthy volunteers (20–35 years; 12 females and 14 males)
participated in the experiment with written informed consent. Two
subjects were excluded from analysis due to excessive EEG artifacts
and deficient performance in the behavioral task (<60% correct). The
study was approved by the Ethical Committee of the Charité University Hospital Berlin and corresponded to the Human Subjects Guidelines of the Declaration of Helsinki.
Stimuli and Behavioral Task
Vibrotactile stimulation of the left index finger was delivered by a
16-dot piezoelectric Braille display (4 × 4 quadratic matrix, 2.5 mm
spacing) controlled by a programmable stimulator (Piezostimulator,
QuaeroSys, St. Johann, Germany). The pins of the Braille display
were simultaneously driven by a 123-Hz sinusoidal carrier signal,
which was amplitude modulated by a 33-Hz sinus function, creating
the percept of a “flutter” vibration (Romo and Salinas 2003) at 33 Hz.
These carrier and modulation frequencies were chosen to minimize
the risk of artifact contamination at EEG frequencies <90 Hz and to
prevent steady-state evoked responses in the EEG frequency range of
main interest (20–30 Hz). The sound of the tactile stimulation was
masked by white noise (∼80 dB) played via loudspeakers throughout
the entire experiment. None of the participants reported hearing any
sound attributable to the tactile stimulation.
An illustration of the experimental protocol is shown in Figure 1A.
After a variable prestimulus interval (1000–1500 ms), each trial started
2230 WM Coding of Analog Stimulus Properties
•
Spitzer et al.
Spectral Analysis
Time frequency (TF) representations of spectral power (5–48 Hz) in
the retention interval (−800–2500 ms, relative to cue onset) were obtained by applying a tapered sliding window fast Fourier transform
(FFT) with a Hanning taper and an adaptive time window of 7 cycles
length, using the FieldTrip function ft_specest_mtmconvol.m provided with SPM8. This method is very similar to a Morlet wavelet
transform (cf. Bruns 2004), but provides additional tapering options
for optimum sensitivity at higher frequency ranges. Here, exploratory
analysis of higher frequency bands (>48 Hz), using a multitapered
FFT, yielded no significant effects. To avoid potential results distortions by systematic stimulus-induced effects in the precue interval
(Supplementary Fig. 2), the TF spectra were normalized relative to the
average power in the entire data epoch (−800–2500 ms) to represent
percentage power changes over time. This normalization approach
retains condition-specific differences in the TF spectras’ temporal
profile only. Control analyses of the average reference spectra used
for the normalization showed no systematic differences in any of the
reported contrasts (all frequency bins P > 0.05), rendering it unlikely
that stationary effects had been masked out in the normalized TF
data. Note that subsequent analysis involved direct comparisons of
the cue-induced TF profiles between the A- and B-tasks after identical
s1 stimulation and encoding conditions (cf. Figs 1 and 2), ruling out a
potential contamination of the delay period results by s1-induced
activity and/or stimulation artifacts. TF bins for which the sliding FFT
window overlapped with s2 onset, at the end of the analysis window,
were excluded from analysis (cf. transparent masks in Figs 1 and 2).
beta parameter estimates for each model regressor at each TF bin. TF
contrasts of interest were then computed by weighted summation of
individual regressors’ beta estimates (cf. Supplementary Fig. 1B–D).
The individual contrast spectra were subjected to mass univariate
analysis on the group level, using 1-sample t-tests as implemented in
SPM8 (for details, see Kiebel et al. 2005; Litvak et al. 2011). This involved computing for each TF bin a t-value that reflected the significance of the contrast. Family-wise errors (FWEs) in TF space were
controlled using random field theory (RFT; Worsley et al. 1996; Kilner
et al. 2005) to determine the FWE-corrected probability that, at a
given channel, a cluster of adjacent significant TF bins may have been
obtained by chance. A cluster was thereby defined as a group of adjacent TF bins that all exceeded a threshold of P < 0.005 (uncorrected).
RFT approximates the probability that a multidimensional statistical
map exceeds some height or extent by chance, under the assumption
that the error terms conform to a smooth random field with a Gaussian distribution (Brett et al. 2003). Convolving the normalized data
with a Gaussian smoothing kernel prior to analysis ensures this assumption. The FWE-corrected significance of a given cluster (Pcluster)
is then determined using the distribution of the expected Euler
characteristic, given the smoothness of the map, under the null
hypothesis of a continuous random field. Analysis of parametric TF
contrasts (Fig. 2, cf. Supplementary Fig. 1C,D) a priori focused on
right prefrontal channels near electrode F2, where parametric modulations of upper beta activity were most consistently observed in our
previous vibrotactile frequency work. If significant (Pcluster < 0.05,
FWE) cue-induced parametric modulations were identified at F2, and/
or at least 2 immediately neighboring right prefrontal channels, we
continued with analyses of the condition-specific modulations in the
respective frequency band, averaged over those adjacent channels
showing a similar effect (cf. Fig. 2). Unless noted otherwise, only
correct discrimination trials were included in the reported analyses.
Where applicable, error trials were analyzed separately.
Figure 2. Parametric modulations of prefrontal oscillatory activity by the
task-relevant stimulus attribute. (A) Modulations by stimulus intensity. Upper left:
Time frequency statistical parametric map of the difference in parametric modulation
by s1 intensity during retention in the A-cue condition compared with the B-cue
condition, for representative right prefrontal channels outlined by dashed rectangle in
upper right. Upper right: Descriptive topographical map (color scaling as in left) of the
parametric modulation outlined by dashed rectangle in upper left. Lower left:
Time-courses of the parametric modulations by s1 intensity in the 2 task conditions,
for the frequency band and channels outlined in upper left. “MI”: maintain intensity
(A-cue); “MD”: maintain duration (B-cue). Lower right: 3D source reconstruction of
the parametric modulation by intensity in the A-cue condition, for the TF window
outlined in upper left. (B) Analogous to A, for the modulation by s1 duration in the
B-cue condition compared with the A-cue condition. Source reconstruction failed to
produce statistically reliable results for the TF window outlined in upper left. A
descriptive illustration of the reconstructed prefrontal source is provided in
Supplementary Figure 3.
Statistical Analysis
The statistical design was implemented in SPM8, using the general
linear model (GLM) applied to each subject’s single-trial TF data. To
warrant conformity with the GLM under normal error assumptions
(Kiebel et al. 2005), the TF data were convolved with a 3 Hz × 500 ms
(full-width at half maximum) Gaussian kernel. The GLM design
matrix (cf. Supplementary Fig. 1A) consisted of 2 dummy variables
specifying the trials’ cue condition (A or B), 2 parametric regressors
coding the intensity of s1 under the respective cue conditions, and 2
parametric regressors analogously specifying the duration of s1. The
vectors coding intensity and duration were zero-centered and normalized. Because across trials, intensity and duration were varied independently of each other, the resulting parametric regressors were
orthogonal, allowing us to estimate their respective impact within the
same regression model. The model was inverted using restricted
maximum-likelihood estimation as implemented in SPM8, yielding
Source Reconstruction
The sources of EEG activity were modeled using 3-dimensional (3D)
source reconstruction as implemented in SPM8 (Friston et al. 2006).
For each participant, a forward model was constructed, using an 8196
vertex template cortical mesh coregistered to the individual electrode
positions via 3 fiducial markers. The forward model’s lead field was
computed using the 3-shell boundary element method EEG head
model available in SPM8 (Phillips et al. 2007). Prior to model inversion, the data were band-pass filtered around the frequency band of
interest. Source estimates were then computed on the canonical mesh
using multiple sparse priors (Friston et al. 2008) under group constraints (Litvak and Friston 2008), including the data from all conditions of interest. For TF windows where significant effects were
identified in the channel-level analysis, condition-specific TF contrasts
were used to summarize oscillatory source power for specific frequency bands and at specific times as 3D images. This entailed convolving the single-trial source activity with a series of Morlet
projectors and weighting the average power from each condition with
a Gaussian window centered on the time interval of interest. The
source reconstructions were then analyzed on the group level using
conventional statistical parametric mapping procedures, using a
display threshold of P < 0.001 (uncorrected). Additional nonsignificant source reconstruction results are illustrated descriptively in Supplementary Figure 3. The SPM anatomy toolbox (Eickhoff et al. 2005)
was used to establish cytoarchitectonic reference where possible.
Results
Behavioral Results
On average, participants correctly discriminated 69.3% of the
stimulus intensities in the A-task, compared with 77.6% of the
stimulus durations in the B-task (P < 0.001, 2-sided paired
t-test). A detailed description of the behavioral data is given in
Table 1. As expected by Weber’s law (Fechner 1966),
Cerebral Cortex August 2014, V 24 N 8 2231
was attributable to traditional beta-band activity (15–25 Hz) in
SI (for details, see Supplementary Fig. 2A).
Table 1
Behavioral results
Accuracy (%)
s1 intensity (a.u.)
A (maintain intensity)
B (maintain duration)
s1 duration (ms)
A (maintain intensity)
B (maintain duration)
RT (ms)
s1 intensity (a.u.)
A (maintain intensity)
B (maintain duration)
s1 duration (ms)
A (maintain intensity)
B (maintain duration)
1600
74.1
75.0
350
69.2
74.0
2200
76.9
78.3
450
69.7
80.9
2800
66.1
77.7
550
67.9
80.1
3400
60.1
79.1
650
69.9
75.4
1600
660
754
350
743
794
2200
671
726
450
691
770
2800
660
746
550
603
730
3400
632
758
650
592
688
Linear slope
−0.53*
0.12
0.00
0.04
Linear slope
−9.5
3.2
−54.1*
−35.8*
Note: Discrimination performance. Upper: Mean s1–s2 discrimination accuracy (% correct) in the
A- and B-tasks for each s1 intensity and duration, and linear regression slopes. Asterisks indicate
significance of a linear trend (Bonferroni corrected). Lower: Same as upper, for RTs (ms),
measured relative to the offset of s2.
discrimination accuracy in the A-task decreased with increasing s1 stimulus intensity (Table 1, upper; P < 0.001, linear
trend analysis, Bonferroni corrected). In contrast, accuracy in
the B-task was not systematically affected by s1 duration
(P > 0.70). Thus, the lower mean accuracy in the A-task compared with the B-task was in particular due to poor discrimination of high s1 intensities, compared with long s1 durations
(cf. Table 1). In neither task did accuracy covary with variations in the uncued stimulus dimension (i.e., duration in the
A-task, or intensity in the B-task, both P’s > 0.05).
Average response times (RTs; measured relative to s2
offset) were longer in the B-task (745 ms) when compared
with the A-task (656 ms; P < 0.001). This is expected because
duration judgments (B-task) require evaluation of s2 in full
length. Likewise, RTs in both tasks decreased with increasing
stimulus duration (Table 1, lower; both P’s < 0.001), but
remained stable across stimulus intensities (both P’s > 0.50).
Oscillatory EEG Responses
First, we explored possible differences in overall cue-induced
TF activity between the 2 task conditions during cued retention (0–2500 ms, Fig. 1B, left). Statistical analysis indicated
that, between 700 and 1300 ms after cue onset, activity in the
traditional sensorimotor beta band (15–25 Hz) over right
centro-parietal channels (C4, C6, CP4, and CP6) was significantly decreased in the A-task compared with the B-task
(Pcluster < 0.05, FWE; maximum at channel CP6, 20 Hz, 1050
ms, t = 4.26). SPM 3D source reconstruction attributed this
effect to the right primary somatosensory cortex (SI, areas 3b,
1, and 2; Fig. 1B, right; P < 0.001) contralateral to s1 application (for related lateralization analyses, see Spitzer and
Blankenburg 2011). No further differences in overall oscillatory activity during the retention interval were found
between the 2 tasks.
Also evident from the time-courses illustrated in Figure 1B,
lower; in both memory tasks, sensorimotor beta activity was
by tendency decreasing throughout the retention interval, in
line with previous findings that tactile WM processing per se
is typically accompanied by suppression of beta activity in
early somatosensory processing areas (e.g., Spitzer and Blankenburg 2011, 2012). Supplementary inspection of this effect
in the present data affirmed that the overall power decrease
2232 WM Coding of Analog Stimulus Properties
•
Spitzer et al.
Stimulus-Dependent Parametric Modulations of
Prefrontal Beta Activity
To examine parametric modulations of oscillatory activity as a
function of the stimulus attribute processed in WM, we analyzed parametric contrasts reflecting the strength of a linear
relation between TF activity and the different properties of s1
(i.e., intensity or frequency, cf. Supplementary Fig. 1). Based
on our previous work, we focused on right prefrontal channels near electrode F2, where delayed parametric modulations
in the upper beta frequency range (20–30 Hz) have been consistently found during WM maintenance of vibrotactile frequency (Spitzer et al. 2010; Spitzer and Blankenburg 2012,
see Spitzer and Blankenburg 2011 for evidence that this type
of activity is typically right lateralized).
We first inspected the cue-induced parametric modulation
by intensity in the A-task, compared with the modulation
by intensity in the B-task (Fig. 2A). This contrast revealed a
significant (Pcluster < 0.05, FWE) cue-induced parametric
modulation in the upper beta frequency range (20–25 Hz)
between 250 and 750 ms after cue onset (maximum at
channel F4, 22 Hz, 400 ms, t = 3.88). Note that the right prefrontal scalp topography of this effect (Fig. 2A, upper right)
might have been reinforced by a priori channel selection.
Figure 2A (lower) illustrates the condition-specific timecourses of the average parametric upper beta-band modulations by s1 intensity. In the A-cue condition, when subjects
retrospectively focused on s1 intensity, a sustained modulation by intensity was evident between 250 and 1200 ms
after cue onset (all time bins P < 0.05). No evidence for a
modulation by s1 intensity was found in the B-cue condition
(all time bins P > 0.05). Source reconstruction (Fig. 2A, lower
right) attributed the parametric modulation by intensity in
the A-cue condition to the inferior frontal gyrus (IFG, area
45, P < 0.001) in the right lateral PFC, which conforms with
previous localizations of frequency-dependent WM activity
(e.g., Spitzer et al. 2010).
Analogously, we examined the cue-induced parametric
modulation by duration in the B-task, compared with the modulation by duration in the A-task (Fig. 2B). This contrast revealed
a significant (P < 0.05, FWE) cue-induced parametric modulation that was spectrally and topographically similar to the
modulation by intensity reported above (23–29 Hz; maximum
at channel F2, 27 Hz, 1550 ms, t = 3.26), but emerged in a later
time window (1300–1650 ms). The condition-specific timecourses of the average parametric modulations indicate that
upper beta was modulated by s1’s duration in the B-cue condition (Fig. 2B, lower, 1300–1700 ms, all time bins P < 0.05)
where subjects retrospectively focused on duration, but not in
the A-cue condition (all time bins P > 0.05), where subjects
focused on intensity. Source analysis of the modulation by duration in the B-task yielded no results that exceeded the display
threshold. Below threshold, at a considerably reduced level of
significance, the effect was attributed to a very similar source as
the modulation by intensity (IFG, area 45, see Supplementary
Fig. 3).
A descriptive summary of the grand-average changes in
upper beta activity as identified in the above analyses is
shown in Figure 3A. The linearity of the group averaged data
Figure 3. (A) Summary illustration of grand-average cue-induced changes in prefrontal upper beta activity as a function of s1 intensity (left) and as a function of s1 duration
(right). (B) Performance-related differences. Left: Parametric modulation by s1 intensity during the maintenance of intensity, computed from correct trials (saturated red)
compared with incorrect trials (light red). On incorrect trials, the parametric modulation was significantly reduced (z = 3.59, P < 0.001; resampling statistics based on 1000
iterations). Right: Same as left, showing a similar trend for the parametric modulation by s1 duration during the maintenance of duration (z = 1.72, P < 0.10).
points may only approximately mirror the group statistical
analysis of individual subject linearity derived from single-trial
GLM analysis (cf. Fig. 2) and is shown for illustrative purpose
only.
showed a trend in the same direction for the duration task
(Fig. 3B, right, 0.0204 vs. −0.0019, z = 1.72, P < 0.10), overall
indicating a reduction in the stimulus-dependent parametric
modulations on error trials, compared with correct trials.
Parametric Modulations on Error Trials
Concluding our analysis, we examined the potential behavioral relevance of the parametric modulation effects reported
above by examining the modulations on correct discrimination trials when compared with error trials. To account for
the smaller number of incorrect than correct trials, we used
resampling statistics (Voytek et al. 2010; Spitzer and Blankenburg 2012) to estimate a surrogate distribution of regression
coefficients computed from subsets of randomly drawn
correct trials (matched to the number of error trials available
per cue and stimulus conditions). For computational efficiency, the resampling analysis was implemented using
simple regression analysis instead of the full GLM model used
in the main analysis (cf. Supplementary Fig. 1).
Applied to correct trials, single-trial regression replicated
the cue-specific modulations by intensity and duration outlined in Figure 2A,B, lower, respectively (mean t-values 2.97
and 2.51; both P’s < 0.05). Applied to error trials, in contrast,
the analysis showed neither a modulation by intensity in the
A condition (t = −0.37; P > 0.70), nor a modulation by duration
in the B condition (t = −0.15; P > 0.80). Resampling statistics
based on 1000 iterations indicated that, relative to the surrogate distribution’s mean, the observed regression coefficient
on incorrect trials was significantly decreased in the intensity
task (Fig. 3B, left; 0.0119 vs. −0.0033; z = 3.59, P < 0.001), and
Discussion
The presented findings provide an important extension of
previous evidence for parametric WM processing of periodic
stimulus frequencies (Romo et al. 1999; Barak et al. 2010;
Spitzer et al. 2010; Spitzer and Blankenburg 2011, 2012), by
demonstrating that oscillatory delay activity in the PFC can be
parametrically driven by other stimulus attributes, as well. In
particular, during retention of either intensity- or duration
information, we identified stimulus-dependent modulations of
similar spectral profiles, with maximum modulation in the
upper beta frequency range (20–30 Hz), thereby strongly resembling the frequency-dependent parametric WM effects reported recently (Spitzer et al. 2010; Spitzer and Blankenburg
2011, 2012). Like in the previous work, the present parametric EEG modulations were found to relate to successful
DMTS task performance and to critically depend on task
demands to actively focus on a singular sensory quantity in
WM. In particular, the observation of spectrally and topographically similar prefrontal modulations by stationary and nonstationary stimulus attributes indicates that the previously
reported frequency-dependent parametric WM phenomena
may not be uniquely linked to any specific temporal or nontemporal properties of periodic frequencies. Together, the
present and previous findings corroborate a view of parametric WM processing in the PFC as a generalized mechanism
Cerebral Cortex August 2014, V 24 N 8 2233
for internal representation of abstract quantity information,
which may be derived for manifold perceptual dimensions
and/or sensory modalities (cf. Spitzer and Blankenburg
2012).
Although the parametric modulations by intensity and duration were spectrally and topographically similar, the intensity
effect showed a slightly more anterior prefrontal scalp distribution, and by tendency covered lower frequencies within the
upper beta range (20–30 Hz). Due to limitations in the spatial
precision of noninvasive EEG recordings, we cannot conclusively infer whether the minor differences on the scalp level
may represent different reflections of the same underlying
cortical process, or may reflect a similar encoding of the
2 stimulus attributes in adjacent prefrontal areas, which
may have been selectively engaged depending on the cue. At
least suggestively, 3D source localization attributed the
present prefrontal effects to the IFG in the lateral PFC, which
conforms very well with previous source localizations of
upper beta modulations during periodic frequency maintenance (Spitzer et al. 2010; Spitzer and Blankenburg 2011,
2012) and with the homolog area found to exhibit quantitative WM coding in monkeys’ lateral PFC (Romo et al. 1999;
Barak et al. 2010).
A prominent difference between the oscillatory representations of the 2 stimulus attributes was their differential timecourse during cued retention, with the modulation by intensity
arising considerably earlier than the modulation by duration.
Notably, very similar time-courses of upper beta modulations
were observed in our previous retro-cue study of vibrotactile
frequencies, where a quantitative WM representation of more
difficult to-be-maintained information arose very early (at
similar times as the present intensity effect), whereas more
easily remembered information was parametrically represented
only after a delay (at similar times as the present duration
effect; cf. Fig. 3. in Spitzer and Blankenburg 2011). This
pattern is paralleled by our present findings, where s1–s2 discrimination of intensity appeared more difficult than discrimination of the stimuli’s duration (cf. Behavioral Results). We
may speculate that, in the retro-cue paradigm, subjects may
have put greater emphasis a priori on the more difficult of the
2 stimulus attributes (here: intensity), which may have facilitated an early prefrontal representation of the to-be-maintained
information. To test this possibility in a post hoc analysis, we
examined the brain–behavior correlation between the early
cue-induced modulation by intensity (as identified in Fig. 2A,
left) and the individual difference in behavioral A- versus
B- task performance, which can be seen as a coarse indicator
of the extent to which individual subjects may have prioritized
intensity- over duration-encoding. Indeed, a positive correlation was found (RPearson = 0.51, P < 0.01), corroborating that
anticipatory prioritization may have contributed to the relatively early onset of the prefrontal representation of intensity
information.
Independent from the stimulus-dependent modulations of
upper beta (20–30 Hz) in PFC, traditional beta-band power
(15–25 Hz) in somatosensory areas was overall decreasing
throughout the retention interval (cf. Fig. 1A, lower and Supplementary Fig. 2A), in line with previous findings that tactile
WM processing per se is accompanied by suppression of sensorimotor beta oscillations (e.g., Spitzer and Blankenburg
2011, 2012). Here, the suppression in SI was found to be transiently reduced during the maintenance of duration, compared
2234 WM Coding of Analog Stimulus Properties
•
Spitzer et al.
with intensity (cf. Fig. 1B). Beta suppression in SI, in the
absence of stimulation, is typically related to tactile attention (e.
g., Bauer et al. 2006; van Ede et al. 2010; Spitzer and Blankenburg 2011). The presented pattern of SI activity thus suggests a
temporary disengagement of tactile attention during retrospective focusing on s1’s duration. Interestingly, this effect was
evident just before the quantitative representation of duration
in the PFC was established (cf. Fig. 1B vs. Fig. 2B), corroborating that quantitative evaluation of the stimuli’s duration involved more abstract, nonsensory processing. Aside from these
task-dependent differences, however, like in our previous
investigations of vibrotactile frequencies, we found no evidence for any delayed parametric representation of the
task-relevant quantitative stimulus attributes in early sensory
areas (for a review of similar findings, see Romo and Salinas
2003; but see Harris et al. 2002; Wang et al. 2012). In reverse,
aside from the specific stimulus-dependent modulations discussed above, right prefrontal upper beta did not reflect overall
(i.e., stimulus-independent) differences between the 2 tasks.
Together, the present results converge with previous findings
that stimulus-dependent prefrontal upper beta modulations are
functionally dissociable from sensorimotor activity in the traditional beta band (Spitzer and Blankenburg 2011, 2012).
Why and how should prefrontal upper beta temporarily increase as a function of the to-be-maintained quantitative
stimulus attribute? One potential interpretation might be in
terms of increased cognitive demands, as might be inferred
from decreased discriminability of greater sensory magnitudes, according to the Weber–Fechner law (Weber 1834).
However, such view of the present findings appears unlikely,
given that s1–s2 discrimination performances declined with
increasing stimulus intensity only, but not with increasing
duration (cf. Behavioral Results). Moreover, stimulusdependent prefrontal modulations during s1 processing were
evident only on correct, but not on incorrect trials, in line
with recent compelling evidence against any direct relation
between right prefrontal beta amplitudes and task demands
in similar WM tasks (Spitzer and Blankenburg 2011, 2012).
An alternative interpretation may be that, in WM, greater
sensory magnitudes might appear more salient. In this light,
modulations of prefrontal upper beta could be seen as form
of mnemonic bottom-up activity, with stimulus-dependent
processing induced by the memory, rather than the actual
presence, of specific sensory input.
In terms of a more mechanistic account for the present
findings, we propose that the prefrontal modulations in upper
beta power are more directly linked to the abstract quantity
information, which is processed in WM. Parametric coding of
analog sensory quantities has been extensively studied in
monkeys (for review, see Romo and Salinas 2003). In particular, during WM maintenance in PFC, such coding was characterized by complex parametric modulations of single cells’
firing rates and by dynamic changes of small population
states over time (Romo et al. 1999; Barak et al. 2010). The
present oscillatory modulations in the human PFC may reflect
a large-scale correlate of such a dynamically changing quantitative representation of the task-relevant WM contents (for
related evidence, see Haegens et al. 2011; Spitzer and Blankenburg 2011). As such, modulations of prefrontal upper beta
may reflect a population-level aspect of the neural code that
represents the task-relevant quantitative information in WM.
We speculate that the parametric (i.e., continuous) nature of
such coding corresponds to an abstract concept of analog
quantity, which might be necessary to perform the subsequent quantitative comparison against s2 (for related discussion, see Nieder and Merten 2007; Verguts 2007).
We recently proposed that parametric prefrontal modulations observed in human EEG may particularly indicate active
updating of WM with the to-be-maintained stimulus attribute’s
quantitative value on an abstract internal scale (Spitzer and
Blankenburg 2011, 2012). In light of such interpretation, an
additional factor may have contributed to the observation that
the modulation by s1 duration occurred relatively late in the
retention interval. Quantitative scaling of s1’s duration may
have been achieved only after a mental replay of the entire
stimulation period (length: 350–650 ms, see Stimuli and Behavioral Task). Thus, in addition to a potential reorientation to the
less-attended stimulus attribute, the retrospective WM processing, or imagery, of time (cf. Coull et al. 2008) may have
further delayed the parametric prefrontal representation by
duration, compared with the representation by s1’s stationary
intensity. In both memory tasks, however, like in the previous
studies of periodic frequency information (Spitzer and Blankenburg 2011, 2012), the parametric EEG modulations were
transient, suggesting that further maintenance of a single quantity value may not necessarily require continuous WM updating
throughout the entire retention period.
In a broader context of previous EEG/MEG investigations of
human WM, mostly in the visual and auditory domains, “parametric” WM processes have typically been assessed by varying
the number of discrete, simultaneously to-be-maintained items
(i.e., the WM “load”). Oscillatory WM effects associated with
load manipulations included parametric modulations of alpha(∼8–13 Hz; e.g., Jensen et al. 2002; Tuladhar et al. 2007), theta(∼4–8 Hz; e.g., Jensen and Tesche 2002), and gamma-band amplitudes (>30 Hz; e.g., Howard et al. 2003; Roux et al. 2012). In
particular, delay activity in the theta- and gamma bands has
been related to the active maintenance of “sensory” representations, with additional phase–amplitude interactions between
the 2 bands supporting the simultaneous representation of
multiple items in WM (e.g., Jensen 2006; Sauseng et al. 2009;
Axmacher et al. 2010). The present quantitative WM task
required the abstraction and maintenance of a single stimulus
attribute only, and focal upper beta modulations in PFC selectively reflected the one task-relevant quantity in the current
focus of attention (for similar results, see Spitzer and Blankenburg 2011). Based on these basic findings, an avenue for
future research can be the integration of analog quantity information on the level of multi-item WM, which may involve more
complex interactions between different frequency bands, and/
or brain areas.
To conclude, we demonstrated parametric WM processing
in the human PFC for different quantitative stimulus attributes, reflected by systematic modulations of upper beta oscillatory activity. The findings extend previous research on WM
maintenance of periodic frequencies and suggest a generalized mechanism for the purposeful updating of abstract quantity information in human WM.
Supplementary Material
Supplementary material can be found at: http://www.cercor.
oxfordjournals.org/.
Funding
This research was supported by a grant from the German
Federal Ministry of Education and Research (BMBF) to F.B.
Notes
We thank R. Auksztulewicz and 2 anonymous reviewers for helpful
comments on the manuscript. Conflict of Interest: None declared.
References
Axmacher N, Henseler MM, Jensen O, Weinreich I, Elger CE, Fell J.
2010. Cross-frequency coupling supports multi-item working
memory in the human hippocampus. Proc Natl Acad Sci USA.
107:3228–3233.
Barak O, Tsodyks M, Romo R. 2010. Neuronal population coding of
parametric working memory. J Neurosci. 30:9424–9430.
Bauer M, Oostenveld R, Peeters M, Fries P. 2006. Tactile spatial attention enhances gamma-band activity in somatosensory cortex and
reduces low-frequency activity in parieto-occipital areas. J Neurosci. 26:490–501.
Benchenane K, Tiesinga PH, Battaglia FP. 2011. Oscillations in the
prefrontal cortex: a gateway to memory and attention. Curr Opin
Neurobiol. 21:475–485.
Brett M, Penny W, Kiebel S. 2003. An introduction to random field
theory. In: Frackowiak RSJ, Friston KJ, Frith CD, Dolan RJ, Price
CJ, Zeki S, Ashburner JT, Penny WD, editors. Human Brain Function II. London (UK): Academic Press.
Bruns A. 2004. Fourier-, Hilbert- and wavelet-based signal analysis:
are they really different approaches? J Neurosci Methods.
137:321–332.
Coull JT, Nazarian B, Vidal F. 2008. Timing, storage, and comparison
of stimulus duration engage discrete anatomical components of a
perceptual timing network. J Cogn Neurosci. 20:2185–2197.
D’Esposito M. 2007. From cognitive to neural models of working
memory. Philos Trans R Soc Lond B Biol Sci. 362:761–772.
Eickhoff SB, Stephan KE, Mohlberg H, Grefkes C, Fink GR, Amunts
K, Zilles K. 2005. A new SPM toolbox for combining probabilistic
cytoarchitectonic maps and functional imaging data. Neuroimage.
25:1325–1335.
Fechner G. 1966. Elements of psychophysics. New York: Holt Rinehart & Winston.
Friston K, Harrison L, Daunizeau J, Kiebel S, Phillips C, Trujillo-Barreto
N, Henson R, Flandin G, Mattout J. 2008. Multiple sparse priors for
the M/EEG inverse problem. Neuroimage. 39:1104–1120.
Friston K, Henson R, Phillips C, Mattout J. 2006. Bayesian estimation
of evoked and induced responses. Hum Brain Mapp. 27:722–735.
Gazzaley A, Nobre AC. 2012. Top-down modulation: bridging selective attention and working memory. Trends Cogn Sci. 16:129–135.
Haegens S, Nácher V, Hernández A, Luna R, Jensen O, Romo R. 2011.
Beta oscillations in the monkey sensorimotor network reflect somatosensory decision making. Proc Natl Acad Sci USA.
108:10708–10713.
Harris JA, Miniussi C, Harris IM, Diamond ME. 2002. Transient
storage of a tactile memory trace in primary somatosensory cortex.
J Neurosci. 22:8720–8725.
Howard MW, Rizzuto DS, Caplan JB, Madsen JR, Lisman J,
Aschenbrenner-Scheibe R, Schulze-Bonhage A, Kahana MJ. 2003.
Gamma oscillations correlate with working memory load in
humans. Cereb Cortex. 13:1369–1374.
Ille N, Berg P, Scherg M. 2002. Artifact correction of the ongoing EEG
using spatial filters based on artifact and brain signal topographies. J Clin Neurophysiol. 19:113–124.
Jensen O. 2006. Maintenance of multiple working memory items by
temporal segmentation. Neuroscience. 139:237–249.
Jensen O, Gelfand J, Kounios J, Lisman JE. 2002. Oscillations in the
alpha band (9–12 Hz) increase with memory load during retention
in a short-term memory task. Cerebr Cortex. 12:877–882.
Cerebral Cortex August 2014, V 24 N 8 2235
Jensen O, Tesche CD. 2002. Frontal theta activity in humans increases
with memory load in a working memory task. Eur J Neurosci.
15:1395–1399.
Kiebel SJ, Tallon-Baudry C, Friston KJ. 2005. Parametric analysis of
oscillatory activity as measured with EEG/MEG. Hum Brain Mapp.
26:170–177.
Kilner JM, Kiebel SJ, Friston KJ. 2005. Applications of random field
theory to electrophysiology. Neurosci Lett. 374:174–178.
Litvak V, Friston K. 2008. Electromagnetic source reconstruction for
group studies. Neuroimage. 42:1490–1498.
Litvak V, Mattout J, Kiebel S, Phillips C, Henson R, Kilner J, Barnes G,
Oostenveld R, Daunizeau J, Flandin G et al. 2011. EEG and MEG
data analysis in SPM8. Comput Intell Neurosci. doi:10.1155/2011/
852961.
Miller BT, D’Esposito M. 2005. Searching for “the top” in top-down
control. Neuron. 48:535–538.
Nee DE, Brown JW, Askren MK, Berman MG, Demiralp E, Krawitz A,
Jonides J. 2012. A meta-analysis of executive components of
working memory. Cereb Cortex. 23:264–282.
Nieder A, Merten K. 2007. A labeled-line code for small and large numerosities in the monkey prefrontal cortex. J Neurosci. 27:5986–5993.
Pasternak T, Greenlee MW. 2005. Working memory in primate
sensory systems. Nat Rev Neurosci. 6:97–107.
Phillips C, Mattout J, Friston K. 2007. Forward models for EEG. In:
Friston K, Ashburner J, Kiebel S, Nichols T, Penny W, editors. Statistical parametric mapping: the analysis of functional brain
images. Amsterdam, Boston: Elsevier/Academic Press.
Romo R, Brody CD, Hernández A, Lemus L. 1999. Neuronal correlates
of parametric working memory in the prefrontal cortex. Nature.
399:470–473.
Romo R, Salinas E. 2003. Flutter discrimination: neural codes, perception, memory and decision making. Nat Rev Neurosci. 4:203–218.
Roux F, Wibral M, Mohr HM, Singer W, Uhlhaas PJ. 2012. Gammaband activity in human prefrontal cortex codes for the number of
relevant items maintained in working memory. J Neurosci.
32:12411–12420.
Sauseng P, Klimesch W, Heise KF, Gruber WR, Holz E, Karim AA,
Glennon M, Gerloff C, Birbaumer N, Hummel FC. 2009. Brain
2236 WM Coding of Analog Stimulus Properties
•
Spitzer et al.
oscillatory substrates of visual short-term memory capacity. Curr
Biol. 19:1846–1852.
Spitzer B, Blankenburg F. 2011. Stimulus-dependent EEG activity reflects internal updating of tactile working memory in humans.
Proc Natl Acad Sci USA. 108:8444–8449.
Spitzer B, Blankenburg F. 2012. Supramodal parametric working
memory processing in humans. J Neurosci. 32:3287–3295.
Spitzer B, Wacker E, Blankenburg F. 2010. Oscillatory correlates of
vibrotactile frequency processing in human working memory. J
Neurosci. 30:4496–4502.
Tallon-Baudry C, Bertrand O, Peronnet F, Pernier J. 1998. Induced
γ-band activity during the delay of a visual short-term memory
task in humans. J Neurosci. 18:4244–4254.
Tuladhar AM, Ter Huurne N, Schoffelen J-M, Maris E, Oostenveld R,
Jensen O. 2007. Parieto-occipital sources account for the increase
in alpha activity with working memory load. Hum Brain Mapp.
28:785–792.
van Ede F, Jensen O, Maris E. 2010. Tactile expectation modulates
pre-stimulus beta-band oscillations in human sensorimotor cortex.
Neuroimage. 51:867–876.
Verguts T. 2007. How to compare two quantities? A computational
model of flutter discrimination. J Cogn Neurosci. 19:409–419.
Voytek B, Davis M, Yago E, Barceló F, Vogel EK, Knight RT. 2010.
Dynamic neuroplasticity after human prefrontal cortex damage.
Neuron. 68:401–408.
Wager TD, Smith EE. 2003. Neuroimaging studies of working
memory: a meta-analysis. Cogn Affect Behav Neurosci. 3:255–274.
Wang L, Li X, Hsiao SS, Bodner M, Lenz F, Zhou Y-D. 2012. Persistent
neuronal firing in primary somatosensory cortex in the absence
of working memory of trial-specific features of the sample
stimuli in a haptic working memory task. J Cogn Neurosci.
24:664–676.
Weber EH. 1834. De Pulsu, resorptione, auditu et tactu: Annotationes
anatomicae et physiologicae. Leipzig: C.F. Koehler.
Worsley KJ, Marrett S, Neelin P, Vandal AC, Friston KJ, Evans AC.
1996. A unified statistical approach for determining significant
signals in images of cerebral activation. Hum Brain Mapp.
4:58–73.