Decoupling Interval Timing and Climbing Neural Activity: A Dissociation between CNV and N1P2 Amplitudes

Abstract

It is often argued that climbing neural activity, as for example reflected by the contingent negative variation (CNV) in the electroencephalogram, is the signature of the subjective experience of time. According to this view, the resolution of the CNV coincides with termination of subjective timing processes. Paradoxically, behavioral data indicate that participants keep track of timing even after the standard interval (SI) has passed. This study addresses whether timing continues after CNV resolution. In Experiment 1, human participants were asked to discriminate time intervals while evoked potentials (EPs) elicited by the sound terminating a comparison interval (CI) were measured. As the amplitude of N1P2 components increases as a function of the temporal distance from the SI, and the latency of the P2 component followed the hazard rate of the CIs, timing processes continue after CNV resolution. Based on a novel experimental paradigm, statistical model comparisons and trial-by-trial analyses, Experiment 2 supports this finding as subjective time is more accurately indexed by the amplitude of early EPs than by CNV amplitude. These results provide the first direct evidence that subjective timing of multisecond intervals does not depend on climbing neural activity as indexed by the CNV and that the subjective experience of time is better reflected by distinct features of post-CI evoked potentials.

Full text

Behavioral/Cognitive
Decoupling Interval Timing and Climbing Neural Activity:
A Dissociation between CNV and N1P2 Amplitudes
Tadeusz W. Kononowicz and Hedderik van Rijn
Experimental Psychology, University of Groningen, 9712 TS Groningen, The Netherlands
It is often argued that climbing neural activity, as for example reflected by the contingent negative variation (CNV) in the electroenceph-
alogram, is the signature of the subjective experience of time. According to this view, the resolution of the CNV coincides with termination
of subjective timing processes. Paradoxically, behavioral data indicate that participants keep track of timing even after the standard
interval (SI) has passed. This study addresses whether timing continues after CNV resolution. In Experiment 1, human participants were
asked to discriminate time intervals while evoked potentials (EPs) elicited by the sound terminating a comparison interval (CI) were
measured. As the amplitude of N1P2 components increases as a function of the temporal distance from the SI, and the latency of the P2
component followed the hazard rate of the CIs, timing processes continue after CNV resolution. Based on a novel experimental paradigm,
statistical model comparisons and trial-by-trial analyses, Experiment 2 supports this finding as subjective time is more accurately
indexed by the amplitude of early EPs than by CNV amplitude. These results provide the first direct evidence that subjective timing of
multisecond intervals does not depend on climbing neural activity as indexed by the CNV and that the subjective experience of time is
better reflected by distinct features of post-CI evoked potentials.
Key words: climbing neural activity; contingent negative variation; interval timing; temporal accumulation
Introduction
A prominent notion in many time perception theories is that
climbing neural activity (CNA; Durstewitz, 2003;Reutimann et
al., 2004; for review, see Merchant et al., 2013a;Wittmann, 2013)
is the primary neural representation of subjective time, with, as a
potential instantiation, the contingent negative variation (CNV),
a slow cortical potential of developing negative polarity at fron-
tocentral scalp locations. Based on correlations between tempo-
ral performance and ramping activity, it has indeed been argued
that temporal integration is indexed by the CNV (Walter et al.,
1964;Macar and Vidal, 2003; but see van Rijn et al., 2011;Wiener
et al., 2012). For example, when participants in temporal gener-
alization studies judge the duration of a comparison interval
(CI), which can be shorter (CI
S
) or longer (CI
L
) than a previously
learned standard interval (SI), a CNV is observed from the onset
of the CI, but this CNA deflects at the offset of the memorized SI
during CI
L
trials (Pfeuty et al., 2005). As no external stimulus is
associated with the deflection, the CNV is suggested to reflect the
neural representation of physical time. However, based on the
CNA hypothesis and the claim that the CNV is the neural repre-
sentation of subjective time, no signatures of temporal sensitivity
should be observed after the CNV starts to deflect. Thus, event-
related potentials (ERPs) triggered by the CI
L
offset should not be
a function of the temporal distance between SI and CI. In contrast
to this view, behavioral data from temporal generalization studies
show that CI
L
accuracy increases with increased distance to the SI
(Wearden, 1992;Ulrich et al., 2006), suggesting that timing pro-
cesses continue after the CNV has resolved.
Interestingly, existing work suggests a link between the ERPs
associated with the offset of a CI and interval timing, as mismatch
negativity (MMN; Na¨a¨ta¨nen and Winkler, 1999) observed in
temporal oddball experiments is a function of the temporal dis-
tance to the SI for short durations (Loveless, 1986;Brannon et al.,
2008).
We contrasted the hypothesis that the end of CNA signals the
end of temporal processing to the hypothesis that interval timing
does not depend on CNA reflected in the CNV by comparing
CI-evoked ERPs. Following the CNA-based hypothesis, the ERPs
to the offset of CI
L
trials should not be sensitive to the distance to
the SI. However, two alternative hypotheses can be extrapolated
from the MMN studies. If the MMN effect is a function of the
distance to the SI, CI-evoked ERP amplitudes should reflect the
distance to the SI. If, however, the MMN effect is driven by aging-
based expectancy (Nobre et al., 2007), the ERPs should decrease
with CI duration, as the expectancy of encountering the offset of
that trial increases with time (Lange, 2009). Figure 1 depicts these
alternative hypotheses.
A second test is presented in Experiment 2. If CNA reflects
subjective time, CNV amplitudes should predict behavioral re-
sponses. Alternatively, if the ERPs to the CIs are driven by tem-
Received June 14, 2013; revised Oct. 16, 2013; accepted Dec. 20, 2013.
Author contributions: T.W.K. and H.v.R. designed research; T.W.K. performed research; T.W.K. and H.v.R. ana-
lyzed data; T.W.K. and H.v.R. wrote the paper.
Thisresearch hasbeen partiallysupported bythe Europeanproject COST(Cooperation inScience andTechnology)
ISCH (Individuals, Societies, Cultures and Health) Action TD0904 “Time In Mental activity: theoretical, behavioral,
bioimaging, and clinical perspectives” (TIMELY; www.timely-cost.eu). We thank Gepke Veenstra, Steffen Bu¨rgers,
and Heleen Meijburg for assistance with data acquisition; and Marc Wittmann for discussions on this work.
Correspondence should be addressed to either Tadeusz W. Kononowicz or Hedderik van Rijn, Experimental
Psychology, Grote Kruistraat 2/1, 9712 TS Groningen, The Netherlands. E-mail: t.w.kononowicz@gmail.com or
hedderik@van-rijn.org.
DOI:10.1523/JNEUROSCI.2523-13.2014
Copyright © 2014 the authors 0270-6474/14/340001-09$15.00/0
The Journal of Neuroscience, February 19, 2014 •34(8):XXXX–XXXX • 1

poral information, the ERP amplitudes might outperform the
CNV in predicting subjective time-based responses.
Materials and Methods
Experiment 1
Participants. Twenty first-year psychology students with no self-reported
hearing loss took part in the experiment and received partial course
credit. Informed consent as approved by the Ethical Committee Psychol-
ogy of the University of Groningen (identification number 11121-E) was
obtained before testing. The data of three participants were not included
in the analyses because of excessive artifacts in ⬎20% of the trials. The
final sample comprised data of 17 participants (all right handed, 6 males)
between 19 and 27 years of age (mean age, 23 years). All cells of the design
contained at least 25 observations.
Stimuli and procedures. A training block familiarized participants with
the standard interval of 2.2 s. It started with five presentations of the
standard interval by means of two tone bursts (50 ms, 500 Hz, ⬃75 dB)
2.2 s apart. Tones were presented at comfortable sound levels using a
Tivoli Audio Model Two system, with speakers located at both sides of
the computer monitor. After these initial presentations, participants
were asked to reproduce the standard interval. During the reproduction
trials, the onset of the interval was indicated by a tone and participants
were asked to press the spacebar (which also triggered the offset tone)
when they thought the standard interval had passed, after which feedback
was provided (Kononowicz and van Rijn, 2011). This training block
continued until the participant reproduced three intervals between 1.98
and 2.42 s (i.e., 2.2 ⫾10%) in a row.
Figure 2 depicts the time course of the experimental trials that followed
the training block. Each experimental trial started with the presentation
of the 2.2 s SI. After an interstimulus interval, sampled from a uniform
distribution from 1 to 2.5 s, the CI was presented, again demarcated by
two tone bursts. Six different comparison intervals were used, each 10%
shorter or 10% longer than the closest neighbor. To reduce noise associ-
ated with response preparation, participants could only respond after a
jittered interval, sampled from a uniform distribution from 3 to 5 s.
Participants were instructed to press the “x” key if they perceived the CI
as shorter than the SI, and the “m” key if the CI was perceived as longer.
No feedback was provided. A gray “⫹” was used as the default fixation
point, replaced by a gray “!” and “x” during the presentation of the SI and
CI, respectively, and by a green “⫹” during the response period.
Trials were presented in randomized order in eight blocks of 24 trials,
with each block containing four repetitions of each CI. Participants re-
ceived feedback after each block, indicating the number of accurate trails
in that block. A short pause (participant paced, 1 min minimum) was
provided between blocks.
EEG acquisition and analysis methods. Electrical brain activity was
measured from 30 scalp locations (Electro-Cap International, tin elec-
trodes: AF3, AFz, AF4, F3, Fz, F4, FC3, FC1, FCz, FC2, FC4, C3, C1, Cz,
C2, C4, CP3, CPz, CP4, P3, P1, Pz, P2, P4, PO3, POz, PO4, O1, Oz, and
O2). To increase the signal-to-noise ratio, all CNV-related analyses are
based on a frontocentral electrode cluster consisting of FC1, FCz, FC2,
C1, Cz, and C2 (Ng et al., 2011). All auditory evoked potential (EP)
analyses focused on FCz, because this electrode showed the strongest
signal in Laplacian topographic plots. Vertical and horizontal EOG ac-
tivity and both mastoids were registered. For all channels, impedances
were kept below 5 k⍀. All channels were amplified and filtered with a
digital finite impulse response filter with a cutoff frequency of 135 Hz
(low pass) using the Refa system (TMS International B.V.) and were
recorded with a sampling rate of 500 Hz using Portilab (TMS Interna-
tional B.V.).
The data were analyzed using FieldTrip (Oostenveld et al., 2011). Off-
line, the signal was referenced to the mastoids, a 50 Hz notch filter was
applied, and data were filtered with a Butterworth filter with a bandpass
of 0.01–100 Hz. Trials containing excessive ocular artifacts, movement
artifacts, or amplifier saturation were excluded from further processing
by visual inspection. Eye blinks, heart beat, and muscle artifacts were
corrected using independent component analysis (Bell and Sejnowski,
1995).
The main focus of Experiment 1 will be the auditory evoked potentials
for the offset tone bursts of the comparison intervals. As we did not
present a CI equal to the SI, no standard MMN could be assessed. In line
with previous work (van Wassenhove et al., 2005;Ng et al., 2011), we
focused on the N1P2 component, defined as the summed absolute am-
plitude of the N1 and P2 peaks. The N1 peak was defined as the minimum
value between 70 and 160 ms after the tone burst, and the P2 peak was
defined as the maximum value between 140 and 300 ms; both ranges
were based on visual inspection of the averaged waveforms.
To filter out CNV-based contamination, a 1–20 Hz Butterworth band-
pass, zero phase-shift filter was applied. As high-pass filtering at 1 Hz may
cause distortions (Acunzo et al., 2012), we also performed all analyses on
0.01 Hz filtered and unfiltered data. As the results were not affected
qualitatively, we will focus on the 1 Hz high-pass-filtered data as the same
filter settings were used by Brannon et al. (2008). After filtering, a Lapla-
cian transformation (local estimate method; Huiskamp, 1991;Oosten-
dorp and van Oosterom, 1996) was applied to improve the spatial
resolution (Nunez and Westdorp, 1994). All trials were baselined to the
average voltage calculated over the 50 ms preceding and following the
onset of the second tone of CI to minimize misalignment of the wave-
forms due to the CNV activity (Correa and Nobre, 2008).
Experiment 2
Participants. Twenty-five first-year psychology students, with no self-
reported hearing loss, took part in the study in exchange for partial
course credit. Informed consent was obtained as for Experiment 1 (iden-
tification number 11191-NE), and data of two participants were ex-
cluded following the rejection criteria of Experiment 1. The final sample
comprised data of 23 participants (all right handed, 6 males) between 19
and 29 years of age (mean age, 23 years).
Stimuli and procedures. A setup similar to that for Experiment 1 was
used (Fig. 1), but with four CI intervals. The S1/L1 durations remained
unchanged, but the S2 and L2 intervals were set to the average of the
S2/S3 and L2/L3 intervals, respectively, resulting in CIs of 1691, 1980,
2420, and 2795 ms. The training block was identical to that in Experi-
ment 1, followed by two blocks of 24 experimental trials from Experi-
ment 1 to familiarize the participants with the general setup of temporal
generalization studies. After these two blocks, the main experimental
phase started.
In this phase, participants were asked to indicate which CI they per-
ceived. Hereto, four gray dots were displayed on the screen after the offset
Figure1. Graphicalsummary ofhypotheses forExperiment1. Theline leftofcenter indicates
that all theories predict an increase in EP amplitudes for CI
S
further from the SI. Predictions for
the CI
L
trials differ. The comparison-based hypothesis, supported by temporal oddball studies
(Brannon et al., 2008), predicts an increase of offset-triggered EP amplitudes as a function of
distance to the SI for the CI
L
trials. Conversely, the aging-based expectancy hypothesis predicts
that EP amplitudes continue to decrease for CI
L
trials as a function of hazard rate (Niemi and
Na¨a¨ta¨nen, 1981;Nobre et al., 2007;Coull, 2009). Both of these hypotheses suggest that timing
continues after the SI duration has passed. The horizontal line depicts the CNA hypothesis. It
predicts no differences in CI
L
-triggered amplitudes, as the CNA assumes that neural integration
takes place until a particular threshold is reached (Durstewitz, 2003;Reutimann et al., 2004;
Simen et al., 2011) and a decision is made. Because a single decision suffices, the CNA hypoth-
esis would predict that no differences are observed for the different CI intervals.
2•J. Neurosci., February 19, 2014 •34(8):XXXX–XXXX Kononowicz and van Rijn •Dissociation between CNV and N1P2 in Interval Timing

of the CI (instead of the green “⫹” of Experiment 1). The first decision
was identical to Experiment 1, as participants had to indicate whether the
perceived duration was shorter or longer than the SI by pressing the “x”
or “m” key. Based on this response, either the left-most or right-most two
dots turned green, and participants were asked to indicate whether the
left or right green dot corresponded to their perception by pressing the
“x” or “m” key. No feedback was provided on these trials.
To obtain a sufficient number of correct and incorrect classified trials,
the number of trials per condition was increased to 56 trials, for a total of
224 trials.
EEG acquisition and analysis methods. The following set of electrodes
was recorded: Fp1, Fp2, AFz, F3, Fz, F4, C3, Cz, C4, P3, Pz, P4, FC1, FCz,
FC2, FC5, FC6, CP5, CP1, CPz, CP2, and CP6. All CNV-related analyses
are based on a frontocentral electrode cluster (Cz, FC1, FCz, FC2) similar
to that used in Experiment 1. The auditory evoked potential analyses
focused on both FCz, because this electrode showed the strongest
signal in Laplacian topographic plots. In addition to these changes,
other elements of the experimental design and EEG recordings were
left unchanged.
Results
Experiment 1
Participants correctly categorized 91% of all trials, ranging from
97% for the most extreme durations (S3, L3) to 82% for the
conditions most similar to the standard duration (S1, L1). Using
a logistic psychometric function, fitted to the proportion of
“long” responses (Wichmann and Hill, 2001;Prins and King-
dom, 2009), the point of subjective equality was estimated at 2236
ms (Fig. 3).
As expected, a CNV was observed during the presentation of
the CIs (Pouthas et al., 2000;Macar and Vidal, 2003;Pfeuty et al.,
2003). Interestingly, the right panel of Figure 4 shows that the
CNV deflects before the SI. To assess the deflection point, we used
the method described by Pfeuty et al. (2003) in which the average
CNV amplitude for successive time windows of 100 ms from
1000 to 2500 ms after sound onset is calculated. The center of the
window with the highest average amplitude determines peak la-
tency. This CNV peak latency was different from 2200 ms for all
three long CIs (L1: 1806 ms, t
(16)
⫽4.06, p⬍0.01; L2: 1700 ms,
t
(16)
⫽4.5, p⬍0.01; L3: 1871 ms, t
(16)
⫽3.3, p⬍0.01). As
expected, because the three long CIs are identical before the SI, a
one-way repeated-measures ANOVA did not show significant
differences among these three conditions (F⬍1). Given the dif-
ference between the CNV peak latency and SI, these results indi-
cate that the deflection of the CNV aligns with the time at which
a binary decision regarding the categorization of the CI interval
can be made, instead of anchored to the memorized SI, as was
suggested in previous work (Macar and Vidal, 2003;Pfeuty et al.,
2005). Note, however, that in previous work the decision could
not be made before the SI because a CI equal to the SI was pre-
Figure 2. Time course of the experimental trials in the temporal comparison task. Both ISI and response delay are sampled from a uniform distribution.
Figure 3. Proportion of long responses per CI length (circles) with a fitted logistic psycho-
metric function (Wichmann and Hill, 2001;Prins and Kingdom, 2009), which is shown in blue.
See Figure 2 for the durations of the CIs.
Kononowicz and van Rijn •Dissociation between CNV and N1P2 in Interval Timing J. Neurosci., February 19, 2014 •34(8):XXXX–XXXX •3

sented. Therefore, a binary shorter/longer
decision cannot be made until around the
time of the SI.
The analyses of the auditory evoked
potentials focus on FCz because Laplacian-
based topography plots show the strongest
signal coming from this location (Fig. 5B).
The main findings of Experiment 1 are de-
picted in Figure 5. The top panels of Fig-
ure 5 show a clear N1P2 complex, evoked
by the sound demarcating the end of the CI.
A two-way repeated-measures ANOVA
with category (short vs long) and distance
(1–3) as factors showed that the N1P2 am-
plitude increases as a function of the dis-
tance between SI and CI (F
(2,32)
⫽5.70,
p⫽0.008,
␩
2
⫽0.26), with no significant
effects for the main effect of category
(F
(2,32)
⫽1.65, p⬎0.1), or for the inter-
action (F⬍1). Planned post hoc comparisons showed clear N1P2
amplitude modulations for the CIs furthest from the SI when
compared with CIs closest to the SI (t
(16)
⫽2.86, p⫽0.01). Post
hoc comparisons between the middle S2/L2 and the extreme CIs
did not reached significance (S2/L2 vs S1/L1: t
(16)
⫽1.80, p⫽
0.09; S2/L2 vs S3/L3: t
(16)
⫽1.98, p⫽0.06), although numerically
the effects followed the predicted V-shaped pattern. Similarly, the
0.01 Hz filtered data showed the main effect of distance (F
(2,32)
⫽
6.96, p⫽0.003,
␩
2
⫽0.30) and no effect of category (F
(2,32)
⫽
1.91, p⬎0.1), nor the interaction between factors (F
(2,32)
⫽2.62,
p⬎0.05).
As the observed amplitudes are sensitive to the passing of time
even after the deflection of the CNV, these results are at odds with
the prediction that the resolution of CNA signals the end of tem-
poral processing. At the same time, the observed nonlinearity
argues against an aging-based expectancy effect. Instead, the
N1P2 amplitude best matches the pattern predicted by the com-
parison account (Fig. 1;Loveless, 1986;Brannon et al., 2008),
implying that timing processes continue after the SI has been
reached and the CNV has deflected.
Although the V-shaped pattern of N1P2 amplitude argues
against the aging-based expectancy effect, it is possible that this
mounting temporal expectancy (Niemi and Na¨a¨ta¨nen, 1981;No-
bre et al., 2007;Coull, 2009), as the probability of the second tone
steadily increases, can be reflected in other features of ERPs (this
possible dissociation between latency and amplitude has been
proposed earlier; McDonald et al., 2005;Vibell et al., 2007; but see
Seibold et al., 2011). We have therefore also assessed the latency
of the auditory evoked potentials evoked by the offset of CIs. As
can be seen in the lower-right panel of Figure 5, the latency of the
P2 component presents a slightly different picture, with shorter
latencies for CIs associated with higher aging-based expectancy
(F
(5,80)
⫽6.40, p⬍10
⫺4
,
␩
2
⫽0.29; no effect was observed for
N1, F⬍1). This latency effect follows the patterns predicted by
the expectancy account, but in terms of latency instead of ampli-
tude, and is in line with the observation that more attended or
expected stimuli are associated with shortened perceptual pro-
cesses (Spence and Parise, 2010).
As visual inspection of Figure 5 suggests amplitude differences
for the N2 component analogous to the N1P2 effect, especially for
the CI
L
trials, we also tested for the N2 effect in the same way as
the N1P2 amplitude. With N2 peak defined as the minimum
value between 270 and 350 ms, only a main effect of category
reached significance (F
(1,16)
⫽10.23, p⫽0.005,
␩
2
⫽0.39) with
neither the main effect of distance nor the interaction reaching
significance (F⬍1, F⫽1.40 respectively). Because this effect was
not the focus of the current study, we refrain from further
interpretation.
Together, the systematic effects on N1P2 amplitude and P2
latency both before and after the SI suggest that timing processes
influence features of these EPs. Where the P2 latency effect might
be related to the attentional enhancement based on the hazard
rate of elapsed time, the origin of the N1P2 effect is more opaque.
That is, although participants in Experiment 1 were asked to
assess the duration of the perceived intervals, participants in a
study by Brannon et al. (2008) showed very similar effects when
auditory durations were presented while attention was focused
on another task (adult: visual detection task; infants: watching a
silent puppet show). Therefore, the effects observed in both the
study by Brannon et al. (2008) and our study might simply be
caused by a low-level physiological effect based on the objective
differences in duration, which might not be accessible for further
cognitive processing. To assess whether the observed N1P2 mod-
ulation is driven by top-down processes related to the subjective
experience of time, Experiment 2 was conducted.
Experiment 2
ANOVA-based analyses
To assess whether the N1P2 amplitude reflects the subjective ex-
perience of time, the N1P2 and CNV amplitudes for correct trials
have to be compared with those of incorrect trials: if the N1P2
effect is driven by objective duration, correctness should have no
influence on the observed amplitude. On the other hand, if the
N1P2 effect is driven by subjective duration, the amplitude for an
incorrect trial should resemble the N1P2 amplitudes associated
with the CI for which the current trail was mistaken. Similarly,
based on the assumption that CNV-based CNA reflects the sub-
jective experience of time, one has to predict that the CNV am-
plitude for an incorrect trial should resemble the amplitude
associated with the subjective percept. To test these hypotheses,
the design of Experiment 1 was changed so that after making a
longer/shorter decision, participants were subsequently asked to
indicate which of the shorter or longer CIs they perceived.
As in Experiment 1, the CNV (Fig. 6) peaked before the SI for
the CI
L
trials (L1 ⫽1800 vs L2 ⫽1765 ms; not different from each
other, t⬍1; but shorter than SI: t
(22)
⫽3.7, p⬍0.01; t
(22)
⫽5.6,
p⬍0.01). Based on the assumption that the deflection of the
CNV might reflect the internal decision to answer “long,” it could
be that participants initiate a second decision process to support
10 5 0 −5 −10
Short
Time (s)
μV
0123
SI
S3
S2
S1
Long
Time (s)
0123
SI
L1
L2
L3
Figure 4. The CNV time course obtained at the frontocentral electrode cluster during presentation of CIs in Experiment 1,
plottedseparatelyfor CIsshorterand longerthanthe SI.Thevertical dottedline, marked SI,indicates the durationof theSIinterval.
4•J. Neurosci., February 19, 2014 •34(8):XXXX–XXXX Kononowicz and van Rijn •Dissociation between CNV and N1P2 in Interval Timing

the decision whether the current duration is the shorter or the
longer of the two long CIs. As this process would result in a new
duration estimation process, and thus a new boost of CNA, EPs to
the long CIs are potentially contaminated. All subsequent analy-
ses therefore focus on the S1/S2 judgments.
Figure 7 depicts the potentials evoked by the onset of the tone
demarcating the end of the S1 or S2 durations. Visual inspection
of the left panel of Figure 7, showing all trials with correct re-
sponses, shows that the S2 trials are associated with higher am-
plitudes than the S1 trials, matching the amplitude effects found
in Experiment 1. More importantly, the right panel of Figure 7
shows that the incorrect S2 trials—when the participants an-
swered S1—are associated with the lower amplitude, similar to
the correct S1 trials. A two-way repeated-measures ANOVA with
correctness, distance, and electrode as factors supports this cross-
over interaction between distance and correctness (F
(1,22)
⫽
16.60, p⫽0.001,
␩
2
⫽0.43). This interaction suggests that the
N1P2 amplitudes correspond to the subjective experience of
time as S2 trials that are erroneously categorized as “S1 trials” are
associated with an amplitude that is similar to the amplitude
associated with correct S1 trials (and vice
versa for the incorrect S1 trials). The main
effect of correctness reached trend level
(F
(1,22)
⫽4.04, p⫽0.051,
␩
2
⫽0.16), and
none of the other factors reached signifi-
cance (all Fvalues ⬍1).
Single-trial linear mixed-model
regression analyses
However, due to the pseudo-experimental
manipulation of correct and incorrect tri-
als (see also Kononowicz and van Rijn,
2011), the number of observations per cell
is variable. To correct for this potential
bias, we also report analyses based on lin-
ear mixed-effects models (Bagiella et al.,
2000;Pinheiro and Bates, 2000;Gelman
and Hill, 2007;Baayen et al., 2008) that
account for unequal cell counts. As linear
mixed-effects models rely on single items
instead of cell averages, the variables sub-
mitted to analysis were scored on single
trials using the same ranges as those in
previous analysis.
The linear mixed-effects model con-
firms the ANOVA analysis. First, we as-
sessed the importance of interaction
between distance and correctness by
means of formal model comparison. Both
models included a random effect inter-
cept per participant, random slopes for
distance per participant, and N1P2 ampli-
tude as a dependent variable. We con-
structed two linear mixed-effects models,
one including the interaction term be-
tween distance and correctness, and the
other one without such interaction term.
Comparison of these models, by means of
Akaike Information Criterion (AIC) as
well as log-likelihood-based
␹
2
statistics,
showed that the model containing the in-
teraction outperforms the model without
the interaction term (⌬AIC ⫽8;
␹
共i兲
2⫽
9.58, p⬍0.002). The effects of the inter-
cept (
␤
⫽30.47, p⬍10
⫺4
), the distance (
␤
⫽3.79, p⬍10
⫺4
),
and the correctness (
␤
⫽2.76, p⫽0.002) were significant. The
interaction effect of distance and correctness was also significant
(
␤
⫽⫺4.68, p⬍10
⫺4
), confirming that subjective similarity
affects the amplitude of the N1P2 component. These results in-
dicate that the amplitude difference between S1 and S2 trials is
3.79 V/m
2
, but that this effect is negated when an S2 trial is
answered incorrectly (as 4.68 needs to be subtracted for those
trials).
According to the view that the CNV reflects CNA, the subjec-
tive experience of time is a function of the CNV amplitude.
Therefore, a main effect of duration should be observed as well as
a crossover interaction between duration and correctness, as was
found for the N1P2 amplitudes. Both ANOVA and linear mixed-
effects analysis were performed, but no effects reached signifi-
cance in the ANOVA. Similar results were observed for the linear
mixed-effect analyses, as apart from the negative intercept that
reflects the typical CNV negativity (
␤
⫽⫺5.00, p⬍10
⫺4
), none
of the other factors or interactions reached the significance level
Figure 5. EPs triggered by the offset of CI interval. A, The top panels illustrate the post-CI auditory evoked potentials for CIs
shorter and longer than the SI, recorded at FCz. Shaded areas represents measurement areas for amplitude and latency of N1 and
P2 components. The bottom left panel depicts the amplitude of the N1P2 complex, which increases as a function of distance to the
SI. The bottom right panel illustrates the latency decrease of the P2 component as a function of CI length. B, Topographies of N1
(80 –120 ms) and P2 (180–240 ms) peaks, collapsed over all participants and conditions.
Kononowicz and van Rijn •Dissociation between CNV and N1P2 in Interval Timing J. Neurosci., February 19, 2014 •34(8):XXXX–XXXX •5

(p⬎0.1). This matches earlier results re-
ported by Kononowicz and van Rijn
(2011), and contradicts the notion that
CNA, as reflected by the CNV, represents
subjective time.
To further test the relative contribu-
tion of N1P2 and CNV amplitude in pre-
dicting the behavioral responses, we
compared mixed-effect logistic regression
models. All models included a random ef-
fect intercept per participant and random
slopes for distance per participant. We
constructed three linear mixed-effects
models, including either N1P2, CNV, or
both components as predictors, and cor-
rectness as a dependent variable. Compar-
isons of these models, by means of the AIC
as well as log-likelihood-based
␹
2
statis-
tics, showed that the N1P2 model out-
performs the CNV model (⌬AIC ⫽5.4;
␹
共0兲
2⫽5.42, p⬍0.001), the combined
N1P2/CNV model outperforms the CNV
model (⌬AIC ⫽3.8;
␹
共2兲
2⫽7.82, p⫽
0.02), and the additional model complex-
ity of the N1P2/CNV model over the
N1P2 model is not warranted (⌬AIC ⫽
1.6;
␹
共2兲
2⫽2.41, p⬎0.1). All these model
comparisons indicate that the fluctua-
tions in CNV amplitude are worse predic-
tors of the behavioral response than the
N1P2 amplitudes. The estimated log-odds
parameters of the N1P2 model showed
significant effects of the distance (
␤
⫽
1.92, p⬍0.001) and N1P2 amplitude
(
␤
⫽0.01, p⫽0.019), and the interaction
between distance and N1P2 amplitude
(
␤
⫽⫺0.02, p⬍0.01). Figure 8 depicts
the estimated effects for the N1P2 and lack
of an effect for the CNV amplitude. The
increase in N1P2 amplitude leads to an
increase of the odds ratio for a correct response in the case of the
S2 condition, and a decrease in the S1 condition.
The comparison between the model containing the CNV am-
plitude as a predictor against the model without the CNV did not
provide evidence that adding that factor to the model improves
the model fit (⌬AIC ⫽0.2;
␹
共2兲
2⫽2.19, p⬎0.1).
Discussion
According to the information-processing theories of interval
timing (Treisman, 1963;Gibbon, 1977;Gibbon et al., 1984;
Church et al., 1994; for review, see Buhusi and Meck, 2005;van
Wassenhove, 2009;Wittmann and van Wassenhove, 2009), an
accumulator keeps track of the passing of time. Earlier work
(Pouthas et al., 2000;Macar and Vidal, 2003;Praamstra et al.,
2006) has suggested that the CNV is the neural correlate of this
process. However, here we show that higher-quality temporal
information is available than can have been provided by the
CNV, indicating that the accumulation measured by the CNV is
unlikely to act as a source of temporal information. This claim is
supported by empirical results from auditory potentials evoked
by the offset of the comparison duration both before (Experi-
ments 1 and 2) and after (Experiment 1) the standard duration
has passed.
To assess timing-related neural signatures after the CNV has
resolved, we focused on the potentials evoked by the offset tone of
CI in a temporal-generalization paradigm. As the N1P2 ampli-
tude was modulated as a function of temporal distance to the SI,
and also for comparison durations longer than the SI, timing
continued after the resolution of the CNV. As the N1P2 ampli-
tude correlated with the absolute difference between SI and CI
duration, these results suggest that the CI–SI similarity drives the
N1P2 effects (Loveless, 1986;Tse and Penney, 2006,2008;van
Wassenhove and Nagarajan, 2007;Brannon et al., 2008;Roger et
al., 2009) and rules out the alternative hypothesis that differences
in N1P2 amplitude are purely driven by aging-based expectancy
(Nobre et al., 2007). Most importantly, although previous studies
have suggested that memory comparison and decision processes
are completed before the end of a CI longer than the SI (Macar
and Vidal, 2003;Tarantino et al., 2010;Lindbergh and Kieffaber,
2013), these results indicate that participants remain sensitive to
the passing of time, even after the CNV has deflected.
In Experiment 2, we presented participants with a novel tem-
poral generalization paradigm in which a relative shorter-/
longer-than-SI decision and a decision about the absolute
perceived duration (“shortest/longest” or “shorter/longer”) had
to be made. As the N1P2 amplitude outperformed the CNV in
5 0 −5 −10
Short
Time (s)
μV
0123
SI
S2
S1
Long
Time (s)
0123
SI
L1
L2
Figure 6. The CNV time course obtained at the frontocentral electrode cluster during presentation of CIs in Experiment 2,
plottedseparatelyfor CIsshorterand longerthanthe SI.Thevertical dottedline, marked SI,indicates the durationof theSIinterval.
Figure 7. The post-CI EPs for CI
S
trials, measured at the FCz electrode and plotted separately for the presentation of S1 and S2
durations. The left panel depicts the EPs in correctly classified trials (i.e., indicated S1 for an S1 trial, or S2 for an S2 trial); the right
panel depicts the EPs for incorrectly classified trials (i.e., indicated S2 for an S1 trial or vice versa). Shaded areas represent mea-
surement areas for amplitude and latency of N1 and P2 peaks.
6•J. Neurosci., February 19, 2014 •34(8):XXXX–XXXX Kononowicz and van Rijn •Dissociation between CNV and N1P2 in Interval Timing

predicting whether participants correctly identified the actual
durations, the CNV cannot have served as the source of time on
which the behavioral responses (and the N1P2 amplitudes) are
based. Combined with Experiment 1, these results provide evi-
dence that subjective timing of multisecond intervals does not
depend on CNA as indexed by the CNV, and that the subjective
experience of time is better reflected by distinct features of
post-CI evoked potentials.
However, it is rather unlikely that modulation of the evoked
potentials reflects a timing process in and of itself. Instead, it is
more likely that the mechanisms underlying the N1P2 generation
are governed by timing. For example, these effects might be anal-
ogous to mismatch negativity effects driven by a temporal com-
parison of CI to SI. Expectancy is an alternative explanation: as
the task requires participants to be constantly focused on the SI,
the appearance of a tone earlier or later than the SI might evoke a
response based on a mismatch of expectancies. The observed
results could therefore be accounted for by assuming that contin-
uous changes in expectancy attenuate the evoked responses to the
offset tones closer to the SI (Lange, 2009; but see Bendixen et al.,
2009;Todorovic et al., 2011;Wacongne et al., 2011;Todorovic
and de Lange, 2012). This explanation is in line with the predic-
tive coding framework (Rao and Ballard, 1999;Friston, 2005,
2009,2010) that suggests that the brain constantly generates ex-
pectations with regard to upcoming stimuli, based on statistical
patterns in the environment. This “prediction error,” expressed
by enhanced neural activity, reflects the mismatch between top-
down predictions and sensory input. Interestingly, the prediction
error is coded by dopamine neurons in the striatum (Schultz et
al., 1997;Schultz, 2002), which is an important element of the
striatal beat frequency model (Matell and Meck, 2004;Meck et
al., 2008; for an introduction, see van Rijn et al., 2014). According
to this model, striatal neurons continuously compare cortical
patterns with the pattern detected at the time of the reward. Once
the offset of the CI is observed, a dopaminergic burst updates the
corticostriatal connections, and this burst might be enhanced
based on the prediction error (for a link between N1P2 amplitude
and dopaminergic activity in interval timing tasks, see Ng et al.,
2011).
Interestingly, although the amplitudes support comparison-
based explanations, the latencies of the post-CI potentials follow
the pattern predicted by aging-based accounts (Niemi and
Na¨a¨ta¨nen, 1981;Nobre et al., 2007;Coull, 2009). As these laten-
cies shorten as a function of the CI duration, they can be ex-
plained by a speed up of perceptual processing as a function of the
probability of appearance of the offset tone. Such a speedup of
perceptual processing to more expected stimuli is known as the
prior entry effect (for review, see Spence and Parise, 2010;Vang-
kilde et al., 2012). Importantly, as changes in expectancy require
a sense of time, the shortening of latencies for CIs also indicates
that the monitoring of time continues until after the SI, an obser-
vation that is also supported by a right frontal CNV peaking at the
end of the CI, even when a medial frontal CNV has already de-
flected (Pfeuty et al., 2003). Whereas this latter finding can have
multiple explanations, both the V-shaped amplitude attenuation
and the progressive shortening of latency with increased expec-
tancy suggest that timing continues after the SI has been reached,
and thus after the CNV has resolved. Moreover, the results of
Experiment 2 even question the role of the CNV in interval tim-
ing before it has deflected. First, the CNV amplitude before the
tone demarcating the offset of the CI did not distinguish between
correct and incorrect behavioral judgments (note that we also
found no effect for the amplitude of the CNV during the SI on
behavioral judgments; Wiener et al., 2012). Second, model com-
parisons showed that the N1P2 amplitude was a better predictor
of behavioral performance than the CNV. These results support
the notion that instead of the CNV driving temporal perfor-
mance (Pfeuty et al., 2003;Macar et al., 1999,2004), another
source of temporal information is available that provides the sys-
tem with more accurate temporal information than can be de-
rived from the CNV.
Obviously, this does bring up the question of what the CNV
represents. The results reported here nicely align with the obser-
vation of Ng et al. (2011), which suggests that the CNV reflects
more general preparatory processes. In the experiment by Ng et
al. (2011), using a temporal generalization study with much more
widely spaced CIs, the CNV reached a plateau at the duration of
the shortest CI and remained at this level until the SI, after which
it deflected. This observation fits with the notion that the CNV
reflects internal preparatory processes to an upcoming stimulus
(i.e., the short CIs) or to an internal signal, which indicates that a
decision about the behavioral response can be made (i.e., switch
from short to long; Meijering and van Rijn, 2009). This interpre-
tation is analogous to the view that the CNV reflects preparatory
or anticipatory processes (Elbert, 1993;Leuthold et al., 2004; for
review, see Van Boxtel and Bo¨ cker, 2004;Kononowicz and van
Rijn, 2011;Ng et al., 2011;van Rijn et al., 2011;Mento et al.,
2013).
Given that this work questions the role of CNA during timing
as measured by the CNV, and suggests that another source of
temporal information might drive both N1P2 amplitudes and the
CNV, one could question whether observed correlations between
CNA and interval timing actually support the hypothesis that
CNA reflects the accumulation of time. Taking the opposite view;
it could be argued that all CNA observed in temporal contexts is
driven by temporal information instead of being the source of
time. For example, the CNA in certain brain areas could be driven
by information coming from coincident detection in the striatum
(Matell and Meck, 2000,2004;Meck et al., 2008;Jin et al., 2009;
Allman and Meck, 2012;Merchant et al., 2013a,b). However,
further work is needed to elucidate the role of CNA in the sub-
jective experience of time.
In this study, we demonstrated that interval timing does not
require CNA as observed in the CNV, questioning the prominent
ΔAccuracy (Odds ratio)
0.98
0.99
1.00
1.01
1.02
N1P2 CNV
S2 S1 S2 S1
Figure 8. The results of logistic linear mixed-effects model fitted to the N1P2 and CNV data.
Data points and error bars depict the model coefficients and SEs.
Kononowicz and van Rijn •Dissociation between CNV and N1P2 in Interval Timing J. Neurosci., February 19, 2014 •34(8):XXXX–XXXX •7

view that the CNV reflects the accumulator as hypothesized in the
pacemaker-accumulator models of interval timing. Moreover,
we showed that the N1P2 amplitudes for the offset of temporally
relevant tones are a better reflection of internal, subjective time
than the CNV.
The CNV is based on temporal information, but it does not
express timing processes per se (Kononowicz and van Rijn, 2011;
van Rijn et al., 2011). Although CNA is an attractive candidate for
the internal representation of time, one should be cautious not to
overinterpret an observed correlation between CNA and the hy-
pothesized accrual of subjective time, as another source of tem-
poral information might drive the observed patterns.
Notes
Supplementalmaterialfor thisarticle isavailableat https://www.researchgate.net/
publication/237201707_Kononowicz-VanRijn-Appendix?ev⫽prf_pub. To
rule out the possibility that V-shaped modulation of N1P2 amplitude
could have been caused by recovery from the refractory period for CIs
longer than the SI, an additional experiment was run. In this study,
participants heard tones separated by intervals ranging from 1.6 to 2.8 s,
but were not instructed to explicitly time these durations. No effect of
amplitude modulation as a function of interval length was observed. This
material has not been peer reviewed.
References
Acunzo DJ, Mackenzie G, van Rossum MC (2012) Systematic biases in early
ERP and ERF components as a result of high-pass filtering. J Neurosci
Methods 209:212–218. CrossRef Medline
Allman MJ, Meck WH (2012) Pathophysiological distortions in time per-
ception and timed performance. Brain 135:656–677. CrossRef Medline
Baayen RH, Davidson DJ, Bates DM (2008) Mixed-effects modeling with
crossed random effects for subjects and items. J Mem Lang 59:390–412.
CrossRef
Bagiella E, Sloan RP, Heitjan DF (2000) Mixed-effects models in psycho-
physiology. Psychophysiology 37:13–20. CrossRef Medline
Bell AJ, Sejnowski TJ (1995) An information-maximization approach to
blind separation and blind deconvolution. Neural Comput 7:1129 –1159.
CrossRef Medline
Bendixen A, Schro¨ ger E, Winkler I (2009) I heard that coming: event-related
potential evidence for stimulus-driven prediction in the auditory system.
J Neurosci 29:8447–8451. CrossRef Medline
Brannon EM, Libertus ME, Meck WH, Woldorff MG (2008) Electrophysi-
ological measures of time processing in infant and adult brains: Weber’s
law holds. J Cogn Neurosci 20:193–203. CrossRef Medline
Buhusi CV, Meck WH (2005) What makes us tick? Functional and neural
mechanisms of interval timing. Nat Rev Neurosci 6:755–765. CrossRef
Medline
Church RM, Meck WH, Gibbon J (1994) Application of scalar timing the-
ory to individual trials. J Exp Psychol Anim Behav Process 20:135–155.
CrossRef Medline
Correa A, Nobre AC (2008) Neural modulation by regularity and passage of
time. J Neurophysiol 100:1649–1655. CrossRef Medline
Coull JT (2009) Neural substrates of mounting temporal expectation. PLoS
Biol 7:e1000166. CrossRef Medline
Durstewitz D (2003) Self-organizing neural integrator predicts interval
times through climbing activity. J Neurosci 23:5342–5353. Medline
Elbert T (1993) Slow cortical potentials reflect the regulation of cortical
excitability. In: Slow potential changes in the human brain: NATO Sci-
ences Series: A (McCallum WC, Curry SH eds), pp 235–251. New York:
Plenum.
Friston K (2005) A theory of cortical responses. Philos Trans R Soc Lond B
Biol Sci 360:815–836. CrossRef Medline
Friston K (2009) The free-energy principle: a rough guide to the brain?
Trends Cogn Sci 13:293–301. CrossRef Medline
Friston K (2010) The free-energy principle: a unified brain theory? Nat Rev
Neurosci 11:127–138. CrossRef Medline
Gelman A, Hill J (2007) Data analysis using regression and multilevel/hier-
archical models. New York: Cambridge UP.
Gibbon J (1977) Scalar expectancy theory and Weber’s law in animal tim-
ing. Psychol Rev 84:279. CrossRef
Gibbon J, Church RM, Meck WH (1984) Scalar timing in memory. Ann N Y
Acad Sci 423:52–77. CrossRef Medline
Huiskamp G (1991) Difference formulas for the surface Laplacian on a tri-
angulated surface. J Comp Phys 95:477–496. CrossRef
Jin DZ, Fujii N, Graybiel AM (2009) Neural representation of time in
cortico-basal ganglia circuits. Proc Natl Acad Sci U S A 106:19156 –19161.
CrossRef Medline
Kononowicz TW, van Rijn H (2011) Slow potentials in time estimation: the
role of temporal accumulation and habituation. Front Integr Neurosci
5:48. CrossRef Medline
Lange K (2009) Brain correlates of early auditory processing are attenuated
by expectations for time and pitch. Brain Cogn 69:127–137. CrossRef
Medline
Leuthold H, Sommer W, Ulrich R (2004) Preparing for action: inferences
from CNV and LRP. J Psychophysiol 18:77–88. CrossRef
Lindbergh CA, Kieffaber PD (2013) The neural correlates of temporal judg-
ments in the duration bisection task. Neuropsychologia 51:191–196.
CrossRef Medline
Loveless NE (1986) Potentials evoked by temporal deviance. Biol Psychol
22:149–167. CrossRef Medline
Macar F, Vidal F (2003) The CNV peak: an index of decision making and
temporal memory. Psychophysiology 40:950–954. CrossRef Medline
Macar F, Vidal F, Casini L (1999) The supplementary motor area in motor
and sensory timing: evidence from slow brain potential changes. Exp
Brain Res 125:271–280. Medline
Macar F, Anton JL, Bonnet M, Vidal F (2004) Timing functions of the sup-
plementary motor area: an event-related fMRI study. Brain Res Cogn
Brain Res 21:206 –215. CrossRef Medline
Matell MS, Meck WH (2000) Neuropsychological mechanisms of interval
timing behavior. Bioessays 22:94–103. CrossRef Medline
Matell MS, Meck WH (2004) Cortico-striatal circuits and interval timing:
coincidence detection of oscillatory processes. Brain Res Cogn Brain Res
21:139–170. CrossRef Medline
McDonald JJ, Teder-Sa¨leja¨rvi WA, Di Russo F, Hillyard SA (2005) Neural
basis of auditory-induced shifts in visual time-order perception. Nat Neu-
rosci 8:1197–1202. CrossRef Medline
Meck WH, Penney TB, Pouthas V (2008) Cortico-striatal representation of
time in animals and humans. Curr Opin Neurobiol 18:145–152. CrossRef
Medline
Meijering B, van Rijn H (2009) Experimental and computational analyses of
strategy usage in the time-left task. In: Proceedings of the 31st Annual
Meeting of the Cognitive Science Society, pp 1615–1620. Austin, TX:
Cognitive Science Society.
Mento G, Tarantino V, Sarlo M, Bisiacchi PS (2013) Automatic temporal
expectancy: a high-density event-related potential study. PLoS One
8:e62896. CrossRef Medline
Merchant H, Harrington DL, Meck WH (2013a) Neural basis of the percep-
tion and estimation of time. Annu Rev Neurosci 36:313–336. CrossRef
Medline
Merchant H, Pe´rez O, Zarco W, Ga´mez J (2013b) Interval tuning in the
primate medial premotor cortex as a general timing mechanism. J Neu-
rosci 33:9082–9096. CrossRef Medline
Na¨a¨ta¨nen R, Winkler I (1999) The concept of auditory stimulus represen-
tation in cognitive neuroscience. Psychol Bull 125:826–859. CrossRef
Medline
Ng KK, Tobin S, Penney TB (2011) Temporal accumulation and decision
processes in the duration bisection task revealed by contingent negative
variation. Front Integr Neurosci 5:77. CrossRef Medline
Niemi P, Na¨a¨ta¨nen R (1981) Foreperiod and simple reaction time. Psychol
Bull 89:133–162. CrossRef
Nobre A, Correa A, Coull J (2007) The hazards of time. Curr Opin Neuro-
biol 17:465– 470. CrossRef Medline
Nunez PL, Westdorp AF (1994) The surface laplacian, high resolution EEG
and controversies. Brain Topogr 6:221–226. CrossRef Medline
Oostendorp TF, van Oosterom A (1996) The surface Laplacian of the po-
tential: theory and application. IEEE Trans Biomed Eng 43:394–405.
CrossRef Medline
Oostenveld R, Fries P, Maris E, Schoffelen J-M (2011) FieldTrip: open
source software for advanced analysis of MEG, EEG, and invasive electro-
physiological data. Comput Intell Neurosci 2011:156869. CrossRef
Medline
Pfeuty M, Ragot R, Pouthas V (2003) When time is up: CNV time course
8•J. Neurosci., February 19, 2014 •34(8):XXXX–XXXX Kononowicz and van Rijn •Dissociation between CNV and N1P2 in Interval Timing

differentiates the roles of the hemispheres in the discrimination of short
tone durations. Exp Brain Res 151:372–379. CrossRef Medline
Pfeuty M, Ragot R, Pouthas V (2005) Relationship between CNV and tim-
ing of an upcoming event. Neurosci Lett 382:106 –111. CrossRef Medline
Pinheiro JC, Bates DM (2000) Mixed-effects models in S and S-Plus. New
York: Springer.
Pouthas V, Garnero L, Ferrandez AM, Renault B (2000) ERPs and PET anal-
ysis of time perception: spatial and temporal brain mapping during visual
discrimination tasks. Hum Brain Mapp 10:49–60. CrossRef Medline
Praamstra P, Kourtis D, Kwok HF, Oostenveld R (2006) Neurophysiology
of implicit timing in serial choice reaction-time performance. J Neurosci
26:5448–5455. CrossRef Medline
Prins N, Kingdom FA (2009) Palamedes: Matlab routines for analyzing psy-
chophysical data. Available at: http://www.palamedestoolbox.org.
Rao SM, Ballard DH (1999) The evolution of brain activation during tempo-
ral processing. Nat Neurosci 2:79–87. CrossRef Medline
Reutimann J, Yakovlev V, Fusi S, Senn W (2004) Climbing neuronal activity
as an event-based cortical representation of time. J Neurosci 24:3295–
3303. CrossRef Medline
Roger C, Hasbroucq T, Rabat A, Vidal F, Burle B (2009) Neurphysics of
temporal dyscrimination in the rat: a mismatch negativity study. Psycho-
physiology 46:1028 –1032. CrossRef Medline
Schultz W (2002) Getting Formal with Dopamine and Reward. Neuron 36:
241–263. CrossRef Medline
Schultz W, Dayan P, Montague PR (1997) A neural substrate of prediction
and reward. Science 275:1593–1599. CrossRef Medline
Seibold VC, Fiedler A, Rolke B (2011) Temporal attention shortens percep-
tual latency: a temporal prior entry effect. Psychophysiology 48:708 –717.
CrossRef Medline
Simen P, Balci F, de Souza L, Cohen JD, Holmes P (2011) A model of inter-
val timing by neural integration. J Neurosci 31:9238–9253. CrossRef
Medline
Spence C, Parise C (2010) Prior-entry: a review. Conscious Cogn 19:364–
379. CrossRef Medline
Tarantino V, Ehlis AC, Baehne C, Boreatti-Huemmer A, Jacob C, Bisiacchi P,
Fallgatter AJ (2010) The time course of temporal discrimination: an
ERP study. Clin Neurophysiol 121:43–52. CrossRef Medline
Todorovic A, de Lange FP (2012) Repetition suppression and expectation
suppression are dissociable in time in early auditory evoked fields. J Neu-
rosci 32:13389 –13395. CrossRef Medline
Todorovic A, van Ede F, Maris E, de Lange FP (2011) Prior expectation
mediates neural adaptation to repeated sounds in the auditory cortex: an
MEG study. J Neurosci 31:9118–9123. CrossRef Medline
Treisman M (1963) Temporal discrimination and the indifference interval:
implications for a model of the “internal clock.” Psychol Monogr 77:1–31.
Tse CY, Penney TB (2006) Preattentive timing of empty intervals is from
marker offset to onset. Psychophysiology 43:172–179. CrossRef Medline
Tse CY, Penney TB (2008) On the functional fole of temporal and frontal
cortex activation in passive detection of auditory devaince. Neuroimage
41:1462–1470. CrossRef Medline
Ulrich R, Nitschke J, Rammsayer T (2006) Crossmodal temporal discrimi-
nation: assessing the predictions of a general pacemaker-counter model.
Percept Psychophys 68:1140–1152. CrossRef Medline
Van Boxtel GJM, Bo¨cker KBE (2004) Cortical measures of anticipation.
J Psychophysiol 18:61–76. CrossRef
van Rijn H, Kononowicz TW, Meck WH, Ng KK, Penney TB (2011) Con-
tingent negative variation and its relation to time estimation: a theoretical
evaluation. Front Integr Neurosci 5:91. CrossRef Medline
van Rijn H, Gu B-M, Meck WH (2014) Dedicated clock/timing-circuit the-
ories of interval timing. In: Neurobiology of interval timing (Merchant H,
de Lafuente V, eds). New York: Springer.
van Wassenhove V, Grant KW, Poeppel D (2005) Visual speech speeds up
the neural processing of auditory speech. Proc Natl Acad Sci U S A 102:
1181–1186. CrossRef Medline
van Wassenhove V (2009) Minding time in an amodal representational
space. Philos Trans R Soc Lond B Biol Sci 364:1815–1830. CrossRef
Medline
van Wassenhove V, Nagarajan SS (2007) Auditory cortical plasticity in
learning to discriminate modulation rate. J Neurosci 27:2663–2672.
CrossRef Medline
Vangkilde S, Coull JT, Bundesen C (2012) Great expectations: temporal ex-
pectation modulates perceptual processing speed. J Exp Psychol Hum
Percept Perform 38:1183–1191. CrossRef Medline
Vibell J, Klinge C, Zampini M, Spence C, Nobre AC (2007) Temporal order
is coded temporally in the brain: Early event-related potential latency
shifts underlying prior entry in a cross-modal temporal order judgment
task. J Cogn Neurosci 19:109–120. CrossRef Medline
Wacongne C, Labyt E, van Wassenhove V, Bekinschtein T, Naccache L, De-
haene S (2011) Evidence for a hierarchy of predictions and prediction
errors in human cortex. Proc Natl Acad Sci U S A 108:20754–20759.
CrossRef Medline
Walter WG, Cooper R, Aldridge VJ, McCallum WC, Winter AL (1964)
Contingent negative variation: an electric sign of sensori-motor associa-
tion and expectancy in the human brain. Nature 203:380–384. CrossRef
Medline
Wearden JH (1992) Temporal generalization in humans. J Exp Psychol
Anim Behav Process 18:134–144. CrossRef
Wichmann FA, Hill NJ (2001) The psychometric function: I. Fitting, sam-
pling, and goodness of fit. Percept Psychophys 63:1293–1313. CrossRef
Medline
Wiener M, Kliot D, Turkeltaub PE, Hamilton RH, Wolk DA, Coslett HB
(2012) Parietal influence on temporal encoding indexed by simultaneous
transcranial magnetic stimulation and electroencephalography. J Neuro-
sci 32:12258 –12267. CrossRef Medline
Wittmann M (2013) The inner sense of time: how the brain creates a repre-
sentation of duration. Nat Rev Neurosci 14:217–223. CrossRef Medline
Wittmann M, Van Wassenhove V (2009) The experience of time: neural
mechanisms and the interplay of emotion, cognition and embodiment.
Philos Trans R Soc Lond B Biol Sci 364:1809–1813. CrossRef Medline
Kononowicz and van Rijn •Dissociation between CNV and N1P2 in Interval Timing J. Neurosci., February 19, 2014 •34(8):XXXX–XXXX •9