memoire_zarebski_2013

Time perception in Metacontrast
Étudiant David R.L. Zarebski
Directeur Jérôme Sackur
Laboratoire de Sciences Cognitives et Psycholinguistique
Année 2012–2013
Cogmaster

Contents
Acknowledgment 4
Introduction 5
1 Metacontrast, time and Time Perception 7
1.1 Metacontrast as a protocol to investigate the dual stream hypothesis 7
1.2 Dorsal route effect of metacontrast: response times . . . . . . . . . . 8
1.2.1 The Fehrer-Raab Effect . . . . . . . . . . . . . . . . . . . . . . 8
1.2.2 Reinterpretations of the Fehrer-Raab Effect . . . . . . . . . . 9
1.3 Ventral route effect of metacontrast: perceived time . . . . . . . . . . 9
1.3.1 What does the U-curves hide:Reeves 1982 . . . . . . . . . . . 10
1.3.2 The Perceptual Delay Didner and Sperling 1980 . . . . . . . . 10
1.4 Investigating the Fehrer-Raab Effect with the Perceptual Delay . . . 11
2 Material and method 12
2.1 Participants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2 Stimuli . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.3 Time course . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.4 Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3 Results 16
3.1 Subjects’ exclusion from analysis . . . . . . . . . . . . . . . . . . . . 16
3.2 Fehrer-Raab effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.3 Temporal Order Judgements (N=7) . . . . . . . . . . . . . . . . . . . 17
3.3.1 Non-motor blocks . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.3.2 Motor blocks . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.4 Subjective Estimations (N=8) . . . . . . . . . . . . . . . . . . . . . . 23
3.4.1 Non-motor blocks . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.4.2 Motor blocks . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
4 Discussion 25
4.1 Anticipation effect vs Perceptual Delay . . . . . . . . . . . . . . . . . 25
4.1.1 Long, Short SOA and the dimensions of visibility . . . . . . . 28
4.2 TOJ and ESTIM consistency . . . . . . . . . . . . . . . . . . . . . . 28
4.3 Motor interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.3.1 Late developments on perceptual and motor latencies . . . . . 30
4.3.2 Two interacting systems . . . . . . . . . . . . . . . . . . . . . 32
4.3.3 Possible refinements . . . . . . . . . . . . . . . . . . . . . . . 36
Conclusion 36
References 37
Additional graphs 42
2 Cogmaster, year 2012–2013

List of Figures
1 The Ebbinghaus illusion . . . . . . . . . . . . . . . . . . . . . . . . . 5
2 RFI and STI illusions . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3 The Müller-Lyer illusion . . . . . . . . . . . . . . . . . . . . . . . . . 7
4 Target and mask in metacontrast . . . . . . . . . . . . . . . . . . . . 8
5 Fehrer-Raab Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
6 Three accounts on visibility and simultaneity . . . . . . . . . . . . . . 10
7 Metacontrast effect over Toj . . . . . . . . . . . . . . . . . . . . . . 11
8 Trial structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
9 Structure of sessions . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
10 TOJ of discarded subjects . . . . . . . . . . . . . . . . . . . . . . . . 16
11 Response time per subjects . . . . . . . . . . . . . . . . . . . . . . . . 17
12 Temporal Order Judgments . . . . . . . . . . . . . . . . . . . . . . . 18
13 Subjective Estimations . . . . . . . . . . . . . . . . . . . . . . . . . . 23
14 Sound vs Target First . . . . . . . . . . . . . . . . . . . . . . . . . . 26
15 Varieties of metacontrast masking stimuli . . . . . . . . . . . . . . . . 27
16 Exponential effect in Subjective Estimations . . . . . . . . . . . . . . 29
17 The underlying logic of ART . . . . . . . . . . . . . . . . . . . . . . . 30
18 Variations of perceptual and motor latencies . . . . . . . . . . . . . . 30
19 Two criteria model for RT TOJ dissociation . . . . . . . . . . . . . . 32
20 Causal representation of motor/perceptual interactions . . . . . . . . 33
21 Explanations for RT-PSS co variations . . . . . . . . . . . . . . . . . 34
22 TOJ: Linear models of selected subjects . . . . . . . . . . . . . . . . 42
23 TOJ per subjects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
24 estim per subjects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
25 Distributions of RT per subjects . . . . . . . . . . . . . . . . . . . . . 45
List of Tables
1 Factorial design of our investigation . . . . . . . . . . . . . . . . . . . 11
2 Effect of SOA on the mean proportion of Target First . . . . . . . . . 18
3 Points of Subjective Simultaneity vs Response Times . . . . . . . . . 20
4 Summary of models used for analysis . . . . . . . . . . . . . . . . . . 26
D.R.L.Zarebski 3

Acknowledgment
I would like to thank all the members of the LSCP (Laboratoire de Sciences Cog-
nitives et Psycholinguistique) as well as the teachers of the Cogmaster I had the
possibility to discuss with during this year. I would like to thank especially Jérôme
Sackur whose direction allowed me to investigate empirically questions I had only
the possibility to treat distantly as a philosopher of mind.
Among the direct contributors of this thesis, I would like to thank Michel Dutat
for his technical help on the experimental design as well as Anne-Caroline Fievet
and Isabelle Brunet for the selection of participants, Shiri Lev-Ari and Hielke Prins
for their help on R and, finally, Jennifer Lawlor for her watchful proofreading.

Introduction
Introduction: dual path models
Figure 1: The Ebbinghaus illusion
One of the most classical view in theories of vision
states that the visual system is distinguished in two
very different streams: a ventro-phenomenological
path and a dorso-motor one. While physiological
differences between ventral regions – V4, posterior,
central and anterior inferotemporal composed mostly
of small, slow and color sensitive parvo cells – and
dorsal regions of the visual system – V5 composed
mostly of big, fast and color blind magno cells – has
been proposed to implement different functional units of mammalians’ vision a long
time ago (Minkowski 1911), the first modern experimental evidence for functional
dissociations in animals (Mishkin and Ungerleider 1982; Schneider 1969) and humans
(Goodale and Milner 1992) suggest that
the ventral stream – slow with long term memory representations – is involved
in phenomenology and recognitional functions while
the dorsal stream – fast with short term memory representations – is mainly con-
cerned with the perception of motion information of which is directly send to
the motor areas
Dual path model’s of consciousness As noted by the majority of the literature,
distinguishing a ventro-phenomenal stream from a dorso-motor one generally carries
the tacit assumption that the latter is merely unconscious nay inaccessible in a strong
sense. 1
This assumption impact crucially on the experimental methodology in a
way we would like to highlight before presenting the classical experimental findings
against the dissociationnist view.
First of all, it should be emphasised that the first experimental evidence for the
functional dissociation of the ventral and dorsal visual systems were the clinical
cases of, respectively, visual agnosia – impaired capacity to verbally indicate the
orientations of slots consecutive from lesions in the visual ventral areas such as the
Lateral Occipital area (LO) – and optic ataxia – impaired capacity to detect the
orientation of the slot to insert a card in it, consecutive from lesions in dorsal areas
(see Goodale and Milner 1992). In every cases, while a certain function is disturbed,
the other one remains unaltered. Yet, the two cases are not symmetrical inasmuch
as the manifestation of visual agnosia is an introspective one (verbal) while the
evidence for optic ataxia is merely behavioural.
1This distinction between ‘conscious’ perception and ‘unconscious’ action is one of the key ingredients if not
the key conception entertained by the dissociation view (Cardoso-Leite and Gorea 2010:110). See also the vision
for action / vision for perception dichotomy based on the evolutionist account on the emergence of phenomenal
consciousness of Goodale and Westwood 2004.
D.R.L.Zarebski 5

Introduction
In their general structures, experimental investigations of double dissociations
in non-clinical cases –subliminal action priming with backward masking– hold this
conscious/non-conscious dichotomy. In this case, given that the phenomenal and
recognitional aspect of the subject’s experience is an non-altered one, illusions are
used as ventro-phenomenological effect resistance of which is less often tested than
the reciprocal possible influences of illusions on motor tasks. The reason for this
methodological asymmetry is closely related with the fact that, while direct subjec-
tive measures of the illusory effects do not allow for a precise quantification of the
effect of motor tasks on the visual component, the illusory effect on motor tasks al-
lows a more controlled measure of behavioural responses. This issue occurs in most
of the complex tasks derived from Goodale and Milner 1992’s experiments.
Figure 2: RFI and STI illusions: central
components of stimuli do not appear vertical
Non-conservative replications of Goodale and
Milner 1992 Naturally, such a strong inde-
pendence of a motor oriented visual system
from a perceptual one is not indubitable. As
an example, later studies highlighted inter-
actions of visual illusions such as the Ebbing-
haus (Franz et al. 2000; Franz, Scharnowski,
and Gegenfurtner 2005 – see fig.1) or the
Müller-Lyer illusion (Heath, Rival, and Bin-
sted 2004; van Doorn, van der Kamp, and
Savelsbergh 2007 – see fig.3) on grasping
tasks 2
. The very possibility of an interac-
tion seems to depend on the level of com-
plexity of the given illusion. Typically, some
illusions due to contextual effect such as the
Rod-and-Frame Illusion (RFI: Di Lorenzo
and Rock 1982, see fig.2) do not impact on
the orientation of the hand while others, like
the Simultaneous-Tilt Illusion (STI), more
local because probably dependent on horizontal connectivity in V1 (Sengpiel, Sen,
and Blakemore 1997), may send information trough the dorsal path thus interact
(Dyde and Milner 2002). To sum up
It is no longer enough to select an illusion, select a visuomotor task, and
then test whether the former affects the latter. We need to ask first where
the likely locus of the illusion is going to be within the brain. Unless the
illusion operates deep within the ventral stream, it is likely to affect both
dorsal and ventral streams.(Milner and Dyde 2003:11)
2See McIntosh and Schenk 2009 for a review.

Figure 3: The Müller-
Lyer illusion
Taking this advice as a guiding principle, we would like
to suggest that most of the experimental paradigms used to
test whether the dorsal and the ventral path interact suffer
from two limitations, for they i) consist in high level func-
tions (grasping) and illusions (size comparison) ii) and con-
sider interactions from perception to motor action without
wondering about the reciprocal possibility.
1 Metacontrast, time and Time Perception
1.1 Metacontrast as a protocol to investigate the dual stream hypothesis
Metacontrast masking seems to fulfil these specifications for i) it has both perceptual
and motor effects ii) while being also a low level perceptive illusion. By choosing to
focus our attention on low level experimental tasks and illusions, we would like to
place ourselves in the same methodology as more recent investigations on dual paths
hypotheses such as Neumann et al. 1992 or Waszak and Gorea 2004 whose approach
on the issue increased considerably the number of possible models for ventro-dorsal
interactions – see sec.4.3 p.29 for details. In this section, we shall detail some of the
most classical perceptual and motor effects of metacontrast masking before focusing
on a particular perceptual effect related with the subjective timing of the stimuli
–see sec.1.3.2.
Visibility Time and metacontrast entertain complex relations that can been sug-
gested through the history of this perceptual effect. As the phi-phenomenon, the
metacontrast masking appears one of the oldest perceptual effect on visibility –
Exner 1868; McDougall 1904; Sherrington 1897 – though without being isolated as
such and extensively studied before Stigler 1910. While the core assumption of most
of the studies realized before Stigler 1910 consisted in revealing real time perceptions
or isolated visual sensations – see Breitmeyer and Ogmen 2006 pp 5–19 for an ex-
tended historical analysis – the backward masking together with the role of spatial
contrast shrugged off the subjective temporal dimension from the investigation of
this phenomenon.
In a nutshell, the main effect of metacontrast masking has to do with the vis-
ibility of the target hampered by a later, identically centered and non-overlapping
mask – fig.4– which depends on various parameters. Besides the target and mask
durations or the contrast between these stimuli and the background, the effect de-
pends crucially on the objective time between targets and masks onsets (Stimulus
Onset Asynchronies: SOA). Typically, the measured visibility together with the sub-
jective intensity of the target varies as a U- shaped function over SOA. It is thus
D.R.L.Zarebski 7

quite visible when synchronised with the mask or earlier than 150ms but nearly nay
completely invisible around 70-80ms depending on the other parameters 3
.
Target
Mask
Stim
ulus Onset
Asynchrony
Figure 4: Target and mask in
metacontrast
Moreover, it should also be emphasized that,
in addition to the lights vs drawn patterns an-
tagonism –see fig.15 p.27– (Kahneman 1968; Sper-
ling 1964), metacontrast masking occurs for vari-
ous background-stimuli colours combinations though
with different masking efficiency across the colour
spectrum (Bevan, Jonides, and Collyer 1970). Fi-
nally, metacontrast masking is also known to produce
gradual effects in nearly visible conditions for, even
in the cases where the target is detectable, the sub-
jective intensity, sharpness of boundaries (Breitmeyer et al. 2006) and homogeneity
of the texture of the target are affected by the mask – see Sackur 2011 for high-level
processing hypothesis for metacontrast masking based on boundaries and texture
processing.
Despite the profusion of perceptual effects associated with metacontrast masking,
we choose to focus here on an other, less classical, gradual effect on the subjective
occurrence of the target – see sec.1.3.2 for details – rather than the sole visibility.
The reason why lies in the fact that we wanted to compare this temporal effect with
the well known Fehrer-Raab effect –see sec.1.2.
1.2 Dorsal route effect of metacontrast: response times
1.2.1 The Fehrer-Raab Effect
The Fehrer-Raab Effect is one of the most classical experimental evidences for
dual paths models dissociating a dorso-motor visual system from a ventro phe-
nomenological one. The results presented in Fehrer and Raab 1962 suggest that,
even in a strongly masked situation –Targetdur = 5ms,Maskdur = 50ms and
60 ≤ SOA ≤ 80ms– for which the target appears an non-homogeneous shorter
and shapeless flash (in foveal stimulation) or even disappears completely (in periph-
eral stimulation), response times never increase in masked condition thus are still
aligned with the target in the same way as in control situation (unmasked) –see
fig.5.
This result has been interpreted as the manifest independence of the dorsal stream
for a long time provided that motor performance seems independent of visibility.
Given the number of pure replications (Fehrer and Biederman 1962; Harrison and
3More specifically, the type of metacontrast masking function is known to vary according to the ratio of Target
and Mask energies since Kolers 1962. U functions are known to occur in type B metacontrast masking (the more
ordinary) in which the mask is less energetic than the Target. On the contrary, visibility in type A metacontrast
–Mask more energetic than the Target– is known to behave in a linear fashion.

1.3 Ventral route effect of metacontrast: perceived time
Fox 1966; Schiller and Smith 1966) together with conservative extensions (Klotz and
Wolff 1995), we decided to use the Fehrer-Raab Effect as our dorsal route effect.
1.2.2 Reinterpretations of the Fehrer-Raab Effect
Figure 5: Fehrer-Raab Ef-
fect: response times remains
locked on the target no mat-
ter if masked or not
Given the central role of the Fehrer-Raab Effect in most
of the Dissociation View together with the fact that
visibility was not tested in Fehrer and Raab 1962’s origi-
nal paradigm, this effect has been reinterpreted from two
critical angles we should take into account.
Firstly, the fact that response time did not increase
but could decrease in masking has also been interpreted
as the result of an integration process (Proctor and Bern-
stein 1974). Put differently, protocols using control con-
ditions as energetic as the Target Mask combination
rather than the Target alone (less energetic) suggest that
response time are, in fact, aligned on the Mask rather than on the Target which
becomes, from this point of view, comparable with a subliminal cue facilitating
the motor responses (Neumann, Esselmann, and Klotz 1993; Steglich and Neumann
2000).
Secondly, one can wonder whether performing motor responses impact the Tar-
get visibility or if, conversely, motor alignment on the Target in masking conditions
depends on its visibility. Indeed, Fehrer and Raab 1962’s interpretation relies cru-
cially on the idea that visibility varies in a motor context the same way it does in
a non motor context. However, results of within trail successions of RT and Target
detection tasks presented in Waszak and Gorea 2004 suggest that the influence of
the Target on motor responses depends on its perception. 4
1.3 Ventral route effect of metacontrast: perceived time
Interestingly, while most of psychophysical investigations of metacontrast mask-
ing noticed, as Fehrer and Raab 1962; Petry 1978, an effect on the apparent dura-
tion of the target, the temporal aspect of the phenomenological experience induced
by meta-contrast masking on the apparent onset of the target has been left apart
until Neumann 1979.
4The results of Experiment 1 support the view that when backward masking is relatively weak (so that the physical
energy of the masked stimulus at a given d’ is also weak), the impact of the masked stimulus on the motor system
depends on whether or not its internal response exceeds the observer’s perceptual response criterion. Waszak and
Gorea 2004:960
D.R.L.Zarebski 9

1.3.1 What does the U-curves hide:Reeves 1982
Even if not directly interested in phenomenal timing per se, Reeves gave an in-
teresting account on interactions between Temporal synchrony judgments tasks and
visibility tasks. The main issue in Reeves 1982 concerns the very nature of the mech-
anism(s) underlying the modulation of visibility. Opposing three different models
implications of which are sketched on fig.6, Reeves used a dual task-based experi-
ment in order to measure for every single trial the visibility –discrete scale from 1
to 6– and the subjective simultaneity –binary responses distinguishing simultaneous
from successive stimuli.
Figure 6: Visibility func-
tions over SOA: V1 dis-
play, respectively, simultane-
ous (solid line) and successive
(dashes line) trials according
to (a) a single process (b) im-
possible motion or (c) two-
process view.
From Reeves 1982
In a nutshell, while it is known that visibility varies as
a U-shaped function over SOA and that the relation be-
tween simultaneity judgments and SOA is a monotonic
one (Sternberg and Knoll 1973), one can ask whether
visibility and perceived synchronicity interact somehow.
Given the implications on subjective temporality of the
two concurrent models –namely the single process view
(a) distinguishing high spacial frequencies of the target
involved in masking from the low spacial frequencies in-
volved in the Toj (Breitmeyer and Ganz 1976) and the
impossible motion view (b) inspired by Kahneman 1967–
Reeves concluded from his data that the metacontrast
masking function involves two very distinct monotonic
mechanisms – target-mask integration vs target-mask
segregation – competition of which, in a Stroud 1967
fashion, produce the masking effect which is thus maxi-
mal when none wins over the other – see Sackur 2013 for
a modern experimental implication of this dual-process
model.
1.3.2 The Perceptual Delay Didner and Sperling 1980
Inspired by Kahneman 1967’s account of metacontrast masking – in a nutshell, the
Impossible Motion view states that metacontrast involves the same mechanisms as
apparent motions (Fehrer 1965, 1966) but results in a masking instead of a lat-
eral sliding because of the absence of unidirectional clues – Didner and Sperling
1980 proposed a multimodal Temporal Order Judgment based experiment (Toj)
investigating the effect of both metacontrast masking and apparent motion on the
subjective occurrence of the target.
Roughly, the proportion of ”Click First” appears higher in metacontrast and
motion than in the control condition for all different SOA and sound-target intervals

1.4 Investigating the Fehrer-Raab Effect with the Perceptual Delay
Temporal Order Judgments Subjective Estimations
without RT Toj Didner and Sperling 1980 Estim
with RT Toj+RT Estim+RT
Table 1: Factorial design of our investigation
– from -90 (sound first) to 90ms– which drives to the conclusion that the target is
somehow delayed by the energy of the mask. Moreover, this delay seems dependent
to the SOA, for a post-hoc PROBIT analysis (Finney 1971) show that the medians
of the functions that fit the data for a given SOA vary in a gaussian-like fashion
for both metacontrast and apparent motion – see fig.7. Interestingly, Didner and
Sperling 1980 endorsed a classical dual path model suggesting that, together with
the Fehrer-Raab Effect, this perceptual delay should be independent of the motor
response in a go task. Yet this effect of SOA on subjective timing seems highly
variable across subjects and such a prediction as never been fulfilled for these two
classical effects have never been investigated together.
1.4 Investigating the Fehrer-Raab Effect with the Perceptual Delay
Figure 7: Metacontrast effect over Toj as a
function of SOA: each point is the median
of the PROBIT function for the given SOA
and condition (• for apparent motion and ◦
for metacontrast).
From Didner and Sperling 1980
Given that the current research suggests
that motor responses do play a role in
perceptive Toj tasks (Corveleyn, Lopez-
Moliner, and Coello 2012), one might ask
whether motor information could somehow
interact with the retrospective mechanisms
responsible for the perceptual delay. To test
this prediction, we will use the original Did-
ner and Sperling 1980’s experiment as a con-
trol condition for a variant with Motor Re-
sponses (RT). This way, it would be possi-
ble to test separately our ventral (Perceptual
Delay) and dorsal effects (Fehrer-Raab Ef-
fect) and thus see whether they interact in
the particular and low level context of metacontrast masking.
Moreover, most of the experiments on motor perceptual interactions either tested
Response times and Temporal Order Judgments tasks in separated blocks or, as it
has been done since Waszak and Gorea 2004, grouped the two tasks within trial
which do not allow to contrast purely perceptual contexts from mixed context. By
distinguishing blocks performed with motor Responses and blocks realized without
it, we hope to disentangle these two regimes and show how these different method-
ologies impact the interpretation to be given in case of interaction.
D.R.L.Zarebski 11

2 Material and method
Subjective estimations Though it shrugs off potential perturbations induced by a
stroboscopic motion effect between the target and time reference stimulus (Lewis,
Matteson, and Dunlap 1977) of an uni-modal Toj such as Matteson and Flaherty
1976’s, the multimodality of Didner and Sperling 1980’ Toj task raises also some
issues, for it is known that the way the system solves the binding problem may
produce strong interaction effects such as the ventriloquism effect of Bertelson and
Aschersleben 2003 – interaction of stereo sound with the horizontal position of the
target in the visual field, see Zampini, Shore, and Spence 2003 for a similar spacial
effects – or the color-tone correlation of Fink et al. 2006. Moreover, Toj being binary
response tasks, an eventual effect can only be seen at the level of the block and not
trial per trial. For both reasons, we choose to use Subjective Estimations of the
temporal occurrence of the target as a complementary protocol.
Despite the fact that temporal estimations of such short durations seem too
difficult and introspective to act as a reliable measure for the Perceptual Delay,
Allan, Kristofferson, and Wiens 1971; Allan and Kristofferson 1974 ’s binary version
of discriminations of small durations (20 vs 30 ms long flashes) seem sure enough
for they follow the Block law. Furthermore, current investigations suggest that
introspective measures of time are reliable in some second order – RT estimation in
Corallo et al. 2008; Gorea, Mamassian, and Cardoso-Leite 2010 – but also in first
order and perceptual tasks – see SOA estimations in Marti et al. 2010 – outside
interference regime (dual tasks conditions).
Given that motor responses constitute such a second task, one can easily imagine
that RT may interfere with subjective estimations of time thus shall we also dis-
tinguish a control condition (without RT) from an interference regime in a similar
way as Toj. Finally, to test Subjective Estimations (Estim) conjointly with Toj
also presents two subsequent advantages for i) it may give a idea of the time scale
concerned with this delay – interval target / sound, target / mask, or larger – and
ii) could, if consistent with Toj, allow for evidence of the Perceptual Delay trial per
trial. Together with the initial dissociation of the Toj task, these considerations
give the factorial design presented in table 1.
2.1 Participants
11 Subjects including the author (DZ), 3 males mean age 24 (standard devia-
tion=3.25), were recruited from an internal list of the LSCP and paid 60 € for
six sessions of about 50 minutes (maximum 2 per day spaced out 2 hours mini-
mum). Ten subjects performed the Temporal Order Judgments experiment. Seven
of whom – DZ, AB, OA, TC, AL, AD, SL – also participated in the Subjective

2.2 Stimuli
Fixation=300ms
ISI=200ms
random=
0 -- 1000ms
Target=25ms
Mask=25ms
ISI=600ms
SOA=
0,30,...150m
s
Fixation=300ms
ISI=200ms
random=
0 -- 1000ms
Target=25ms
Mask=25ms
ISI=600ms
SOA=
0,30,...150m
s
= -90,-60,...,90 ms
Temporal Order Judgements
Subjective Estimations
Figure 8: Trial structure
Estimations (Estim) experiment with an other subject (AC) –see sec.2.4.
2.2 Stimuli
Experiments were coded in Python using the pygame graphical library. Observers
sat at 75cm from a BenQ XL2410T LED monitor with a refresh rate of 120Hz and
a resolution of 1920x1080 px (pitch = 0.272 mm). All stimuli were displayed with
dark (4 cd/m2) pixels on a light grey (23 cd/m2) background. All trials – see. fig.8
– consisted in a fixation cross (0.3°) followed by a target (0.6°) masked (0.8° non
overlapping square) for half of the trials. Target and mask durations were both 25ms.
It should be emphasized that Target was always visible to test Waszak and Gorea
2004’s prediction – see sec.1.2.2 p.9. Sounds used in Temporal Order Judgement
were 1000Hz 5ms long square signals of 60 dB produced with an Arduino Uno on
AKG K512 headphones. Target could appear randomly above or below the fixation
cross (0.86°) in the centre of the x-axis so as to avoid, together with the stereo sound,
any possible ventriloquism effect as described in Bertelson and Aschersleben 2003;
Keetels and Vroomen 2005; Zampini, Shore, and Spence 2003; Zampini et al. 2005.
2.3 Time course
For both Subjective Estimations (Estim) and Temporal Order Judgements (Toj),
trials started with a 300ms long fixation followed by a uniformly distributed random
interval from 200 to 1200ms. Together with the fixation time, delay is thus the 500
to 1500ms interval between the onsets of fixation and target. For the Estim task,
D.R.L.Zarebski 13

subject were told about the uniformity of the distribution.
In half of the trials, the target was followed by a mask with a Stimulus Onset
Asynchrony (SOA) from 0 –i,e, synchronous target and mask– to 150ms by steps
of 30. We added this synchronous apparition of Target and Mask as an equally
energetic control condition to test whether motor responses were aligned on the
former rather than on the latter as this has been suggested by Neumann, Esselmann,
and Klotz 1993; Proctor and Bernstein 1974 – see sec.1.2.2 p.9 for details. For the
Toj task, the interval between the onset of the Target and the onset of the Sound
(∆) was uniformly distributed from -90ms (sound first) to +90ms by steps of 30.
Again, in order to make the task as easy as possible, subject were also told about
the symmetry of ∆ but were not told about the existence of ∆0 –i,e, simultaneous
sounds and targets– in test sessions.
600ms after the end of the target or the mask if masked, subjects performed either
their Subjective Estimations or Temporal Order Judgments without any temporal
constraint except their being told to spend less than 3 seconds for their responses (see
sec.2.4 for details). As soon the answer given, a feedback was displayed for 300ms
in training periods. Inter-trials intervals – grey screen – were uniformly distributed
random delays from 1900 to 2100ms.
2.4 Procedure
Despite the fact that some participants were tested on both tasks, periods did not
overlap.
TOJ For Toj, subjects were asked to report what they perceived first: target or
sound. They were explicitly asked to judge of the order of the onsets and were warned
against any over-intellectualisation of the task. Materially speaking, answers were
given by moving the mouse along the y-axis in order to magnify icons corresponding
to targets and sound and clicking to confirm the choice. As feedback, a red circle
indicated the first event during training trials
ESTIM For Estim, subjects had to place on a linear continuous scale the occur-
rence of the target by moving the mouse along the x-axis and clicking to confirm
their answers. The scale possess some temporal points of reference such as a) a
representation of the fixation – thick line see fig.8 – and b) and bounds indicating
the delay period. Subject were also told that the initial position of the blue cursor
was random and that delays were uniformly distributed random values. For training
trials, corrections consisted in red cursor indicating the real delay.
Training sessions Each subject had to be trained during a specific session mainly
composed with corrected trials. In the first two blocks of these sessions – see fig.9 –

2.4 Procedure
target a) were bigger (1.5°) b) unmasked and, in the case of Toj, c) extreme values
– -120ms, +120ms, half of each – were used as ∆ in order to allow participants to
develop and reinforce an unbiased estimation of time or order.
In the two last blocks, subjects were first trained and corrected on stimuli sets
without synchronous Targets Sounds – 6(SOA)×6(∆)×2 (masked vs unmasked) – to
be put, in a second time, in an experimental situation without feedback on standard
stimuli sets plus 12 trials easier for Toj –∆(−150ms, −120ms, 120ms, 150ms) × 3–
so as they can sometimes feel confident about their judgements.
Motor responses As said in sec.1.4, investigating the relationship between the Per-
ceptual Delay and the Fehrer Raab Effect involve studying the possible interaction
of motor responses with subjective timing tasks. This is the reason why subjects
were asked, for some blocks, to press a response button with the left hand finger as
soon as they detected the target before their Estim (resp. Toj).
Collected by the Arduino uno so as to avoid lags of serial BUS, Response Times
(RT) were then imported with the py-serial library in order correct the subject in
case of anticipation (RT < 50ms) or distraction (RT > 1000ms) by mean of red font
messages in French as ”Trop vite! Vous avez anticipé” (i,e, to fast! You anticipate)
and ”Trop lent! Concentrez-vous” (i,e, To slow! Focus). In training sessions, the 2nd
and 4th blocks were the blocks with motor responses – Estim+RT (resp.Toj+RT).
Training sessions
Test sessions
Unmasked
Unmasked
72 72
Corrected
72
Corrected
72
72+12
Corrected
12 84+12
72+12
84+12
P P P
Corrected
12 84+12 84+12
P P P
Figure 9: Structure of sessions: P stands for
”pause”. Grey rectangles indicate that instruc-
tions precede the given section. Numbers indi-
cates the number of trials.
Test sessions Participants performed
then 5 test sessions per tasks. Ev-
ery test session was decomposed in
two sub-sessions –one with RT, or-
der alternated across sessions– pre-
ceded by 12 corrected trials ran-
domly chosen. Then, subjects per-
formed two shuffled blocks composed
of standard experimental stim-lists –
6(SOA) × 7(∆) × 2 (masked vs un-
masked) – plus 12 trials easier for Toj
–∆(−150ms, −120ms, 120ms, 150ms)×
3.
Each of these two 192 trials-long periods were interrupted by 3 18 seconds-long
pauses during which subjects could have a feedback on their performance during
the immediately preceding period by mean of the proportion of correct and forced
choice situation –though subjects did not know about ∆0– answers for Toj and
the absolute value of the mean of the difference between objective delay and their
estimations for Estims.
D.R.L.Zarebski 15

3 Results
3 Results
Analysis was performed with R version 3.0.1 (”Good Sport”) and the lme4 library
for mixed models.
3.1 Subjects’ exclusion from analysis
−50 0 50
0.00.20.40.60.81.0
CP
delta
0.000473889
−50 0 50
0.00.20.40.60.81.0
EC
delta
0.0014881
−50 0 50
0.00.20.40.60.81.0 HB
delta
0.000467515
−50 0 50
0.00.20.40.60.81.0
JL
delta
0.000356866
click first | ∆ | target first
ProportionofTargetfirst
Figure 10: TOJ of discarded subjects:
prop(Target First) as a function of ∆. Lines
represent linear regressions coefficients of
which appear below
Trials of the training sessions for ESTIM
and the TOJ task were excluded from fur-
ther analysis. Only 8 subjects remained
from the 11 initials. Given the unusual use
of Subjective Estimations in psychophysics,
we choose to select subjects on Temporal
Order Judgments. Three participants –CP,
HB, JL– could not perform the TOJ task
properly enough to avoid floor effects. We
first compared the proportion of Target First
for every single Target-Sound intervals (∆)
with linear regressions. As fig.10 suggest it,
the variability of Target-Sound order did not
make much difference for CP, HB and JL
suggesting that they could not even distin-
guish the order guessing from noisy percep-
tual information. Moreover, we also consid-
ered consistency as a crucial criterion. That is the reason why we excluded EC
coefficient of whom was, given discrepancies, an artificial one –for good subjects see
fig.22 p.42.
3.2 Fehrer-Raab effect
Response times faster than 50ms (0.875%) and slower than 1000ms (0.452%) were
discarded from analysis. We first analysed RT so as to know whether they were
impacted by SOA in this particular context. We may first note that, while RTs
vary greatly among subjects – see fig.11 and fig.25 p.25 for detailed distributions –
masking do not seem to have a clear unidirectional effect nor do SOAs.
We first performed a repeated measure one-way ANOVA with subjects as random
factors on masked trial –excluding the special case of SOA = 0ms– and found no
effect of SOA on the median of RTs (F(4, 28) = 0.569 p > 0.5) and a marginally
significant one on their standard deviations –(F(4, 28) = 2.397 p = 0.074). We
replicated the same analysis distinguishing data according to the perceptual task
realized (TOJ or ESTIM) and did not find any effect of SOA on RT in Temporal

3.3 Temporal Order Judgements (N=7)
Order Judgments (F(4, 24) = 1.289 p = 0.302) nor in Subjective Estimations blocks
(F(4, 28) = 0.408 p = 0.801). Given these results, we performed, then, a paired
T-test between masked and non-Masked result of which shows that the masked
condition does not differ from the unmasked condition –t(7) = −0.588 p > 0.5. In
other words, given that i) masking does not delay RT and that ii) SOA does not
interact with RT in masked conditions, we do have a Fehrer-Raab Effect.
200220240260280300320
soa
MedRT
30 60 90 120 150
suj
AB
AC
AD
AL
DZ
OA
SL
TC
200220240260280300320
soa
MedRT
30 60 90 120 150
suj
AB
AC
AD
AL
DZ
OA
SL
TC
Figure 11: Response time per subjects:
dashed lines stand for unmasked conditions
while solid lines indicate response time as
functions of the SOA
The case of SOA=0ms We kept masked
SOA=0ms for later analysis because of its
special status. As such, the masked condi-
tions with SOA = 0ms (SOA0
) does not con-
sist in metacontrast masking but in some
bigger target (≃ 50%) that could, for ener-
getic reasons, speed up response time. Yet,
a paired T-test through subjects between re-
sponse times in non-masked condition and
this equally energetic control revealed no
difference between these conditions (t(7) =
−0.0087 p > 0.99). In other words, this ab-
sence of merely energetic explanation chal-
lenges the alternative account on the Fehrer-
Raab Effect proposed by Neumann, Essel-
mann, and Klotz 1993; Proctor and Bern-
stein 1974 – see sec.1.2.2 p.9 for details –
inasmuch as it indicates that motor responses are locked on the Target rather than
the Mask.
We first analysed TOJ data which are summarized across subjects in fig.12(a) –
details per subjects can be found in fig.23 p.43. In Temporal Order Judgments the
dependant variable we were interested in was the proportion of Target First re-
sponses across the range of Target Sound intervals (∆). In case of strict simultaneity
of Target’s and Sound’s onsets, the target’s former perception win over the sound’s
more often. This bias seems also dependent on the SOA. For the sake of clarity, we
analysed separately data without motor responses and data with motor responses.
3.3.1 Non-motor blocks
ANOVA based analysis We first tested the influence of SOA with a two-way ANOVA
(∆×SOA) –SOA0
excluded– with subjects as random factors on masked trials only
D.R.L.Zarebski 17

3 Results
0.00.20.40.60.81.0
Without motor responses
−90 −30 30 90
soa
30
60
90
120
150
0.00.20.40.60.81.0
With motor responses
−90 −30 30 90
soa
60
30
120
90
150
proptargetfirst
(a) SOA vs Unmasked
0.00.20.40.60.81.0
−90 −30 30 90
soaC
short
long
no
0.00.20.40.60.81.0
−90 −30 30 90
soaC
short
long
no
click first | ∆ | target firstproptargetfirst
(b) SOA grouped
Figure 12: Temporal Order Judgments averaged across participants: x-axis indicate the interval
between targets and sounds onsets (∆) and y-axis represent the proportion of ”Target first” answers.
Fig.12(a) displays responses for every single SOA while fig.12(b) groups conditions in Long SOA
–120,150 ms– Short SOA –30, 60, 90ms– and Unmasked conditions. Details per subjects can be
found in fig.23 p.43
SOA 30 60 90 120 150
Prop(target) 0.7439 0.6987 0.6453 0.5986 0.5665
Table 2: Effect of SOA on the mean proportion of Target First
(means for every SOA × ∆ × subject combinations). The test revealed i) an ef-
fect of SOA on the prop ofTarget First(F(4, 24) = 7.527 p < 0.005) summarized in
tab.2 ii) but no interaction of SOA with ∆ (F(6, 24) = 0.56 p > 0.9) which exclude
explanations of the effect of SOA based on Sound an Mask contiguity – in case of
interaction, it could have been argued that answers were biased by the fact that the
sound occurs between the Target and the Mask for positive ∆ and long SOA. In
other words, the longer the SOA, the later the Target in perceived. Such a SOA-
induced delay is not a mere energetic effect, for a T-test between the proportion of
Target First in non-masked and SOA0
condition paired by subjects and ∆ revealed
no difference (t(48) = 0.913 p > 0.1). In other words, the modulation of tempo-
rality in this context can be genuinely attributed to metacontrast masking rather
than merely energetic differences as Waszak, Cardoso-Leite, and Gorea 2007 already
suggested it with their own equally energetic control condition (decentered mask).
Though sufficient to test the effect of SOA on this macro level, ANOVA were not
well fitted for later and more complicated analysis. We choose to use Generalized
Linear Mixed Models (GLMM) for two reasons. The first one, pragmatic, came
from the fact that, while we wanted to quantify the Points of Subjective Simultane-

ity (PSS) relative to the different conditions, because of the great variability across
subjects together with the few data points per conditions, we could not perform prob-
ability unit (PROBIT) fit (Finney 1971) for every motor×SOA×subject conditions
without producing locally nonsensical inflexion points or gradients unsuitable for any
between-subjects analysis. The other reason, methodological, came from the way
GLMM deal with individual variability. 5
Among the methodological reasons sug-
gested in favor of Generalized Linear Mixed Model (GLMM) in Moscatelli, Mezzetti,
and Lacquaniti 2012, GLMM is especially preferable over two-levels approaches –or
Parameter-As-Outcome Models (PAOM)– for three reasons:
• GLMM distinguish within and between subjects error terms thus
• take into account the subject-specific standard error
• does not attribute the same weight to every subject as PAOM tacitly does6
Mixed Models analysis We thus choose to use Generalized Linear Mixed Models
(GLMM) for binomial distributions (Target First vs Sound first) with participants’
intercepts as random factors. For the sake of clarity, given the complexity of our
factorial design and the subsequent fact that the same models have been applied on
different subsets of the data, we choose to name explicitly our models and sum up
their predictors, results, domains and interaction in table.4 p.26.
We first started to model (Tmask) the effect of masking on the proportion of
Target First (Delta × Masking). Given the similar effect of SOA=0ms masked
condition and non-masked condition, we conflated the former with the latter so as
to compare these two control conditions with the set of metacontrasted trials. We
observed that metacontrast masking does not produce a Perceptual Delay as in
Didner and Sperling 1980 but, rather, some kind of Anticipation effect for we found
a positive effect of metacontrast masking (β = 0.538 z = 9.13 p < 0.0001) increasing
the proportion of Target First. We did not find any interaction (β = 0.00068 z =
0.673 p > 0.5) between these predictors.
We then took masked trials only – SOA=0ms excluded – to model (Tsoa) the
(Delta × SOA) interaction and found a negative effect of SOA on the proportion
of target first (β = −0.009 z = −8.067 p < 0.0001) and a marginally significant
interaction between these predictors (β = −0.00003 z = −1.757 p = 0.0789). Refor-
mulated, these results imply that i) metacontrast masking produce an Anticipation
effect compared with the control conditions (unmasked and masking SOA = 0ms),
that ii) this Anticipation effect is modulated by SOA duration (important for short
5This point is even more relevant in the case metacontrast masking inasmuch as the existence of different classes
of metacontrast masking observers has been lately suggested in Albrecht, Klapötke, and Mattler 2010; Bachmann
2010.
6As an additional point, the PAOM implicitly assumes that different subjects have the same weight in the second-
level analysis, which is an incorrect assumption when, for example, the number of trials is different from subject to
subject. Moscatelli, Mezzetti, and Lacquaniti 2012:4
D.R.L.Zarebski 19

3 Results
control metacontrast masking
SOA no 0 30 60 90 120 150
PSS no-Motor Responses -17.068 -15.489 -107.571 -76.54 -66.404 -33.63 -19.32
PSS Motor Responses -9.169 -11.788 -122.312 -66.714 -34.556 -29.727 -21.111
RT 260.36 257.50 255.69 255.50 258.94 257.12 258.81
Table 3: Points of Subjective Simultaneity (PSS) vs Response Times (RT): PSS acquired by the
Delta method, RT results from ANOVA’s realized on the medians of RT for the given subset – see
sec.3.2
SOA, decreased with SOA increasing) and that iii) this effect cannot be explained
trivially by the Sound an Mask order variations across SOA (because of the absence
of interaction).
Estimations of Point of Subjective Simultaneity We finally computed the Points
of Subjective Simultaneity (PSS). Again, we did not use a Parameter as Outcome
procedures but a Delta method (Davison 2003:33-35). According to Faraggi, Izikson,
and Reiser 2003, PSS consists in the ratio of two variable parameters (see eq.1)
PSS = −
βO
β1
(1)
for β0 the intercept and β1 the slope fixed effects parameters of a Generalized
Linear Mixed Model. We thus subset our data according to the table 3 to model
(Tdelta: ∆ as unique predictor) individually these subsets with GLMM using PRO-
BIT as link functions to finally perform the Delta Method of estimation with a 0.95
confidence interval implemented in the MERPsychophysics R library developed by
Alessandro Moscatelli – see table 3 for results.
Long and short SOA The effect of SOA on the masking function might not be a
continuous one but posses a step around 80ms as this has been suggested across the
literature on metacontrast masking (Reeves 1982; Sackur 2013) –see sec.4.1.1 p.28
for details. To investigate this possibility, we performed a post-hoc analysis which
consisted in grouping data in two categories according to their SOA – a group of
two vs a group of three and a group of one vs a group of four– using every possible
combinations of SOA to form these groups (15 possible combinations). For every
single partitioning we performed a Generalized Linear Mixed Model fitted by the
Laplace approximation using (∆ × SOAcategorized) as predictors.
Our intuition was as follow: if one partitioning was more natural than an other,
indicator of the goodness of fit such as the Bayesian Information Criterion (BIC)
and the maximum Log Likelihood would be better for this given partitioning –
lower in the case of BIC and higher in the case of the logLik. The partitioning
(30,60) (30,60,90) obtained better values for both BIC and logLike. Though this

methodology do not allow to support strong assumptions about the existence of
such a step, we nonetheless took this result as a relative clue about the fact that
short SOA (30, 60) and long ones (90, 120, 150) do not impact TOJ in the same
fashion.
Adding unmasked trials as a new value for the SOA-categorized variable (Un-
masked, Long, Short SOA), we finally modelled (TsoaC) the influence of (∆×SOA−
categorized) on the proportion of Target First. We found that, compared to long
SOAs, Target were delayed in unmasked conditions (β = −0.323 z = −4.495 p <
0.0001) and anticipated in short soa conditions (β = 0.67 z = 6.667 p < 0.0001). In
accordance with precedent analysis, we did not find any interaction between ∆ and
SOA-categorized.
3.3.2 Motor blocks
To analyse data with motor responses, we first started to performed the same anal-
ysis as in non-motor blocks before measuring the motor effect for itself.
Replication of the analysis for motor blocks We first modelled (Tmask) the (∆ ×
Masking) interaction and found a similar effect of masking on the proportion of Tar-
get First (β = 0.473 z = 8.15 p < 0.0001) and, again, no interaction between these
to predictors (β = −0.00003 z = −0.315 p > 0.5). The effect of SOA on masked
trials was, however, more surprising. Though the model Tsoa (∆ × SOA) revealed
an equivalent effect of SOA in this motor context (β = −0.0063 z = −6.25 p <
0.0001) results also indicated an indubitable interaction between these predictors
(β = −0.00012 z = −7.216 p < 0.0001) which means that the decreasing effect of
SOA on the Proportion ofTarget First becomes stronger as ∆ reach positive values
– i,e. when the Sound comes after the Target.
Then, we performed the same calculation of PSS for every masked and unmasked
conditions Unmasked, SOA0
, SOA30
, ...SOA150
by mean of the same simple one pre-
dictor model (∆) as the one used in non-motor blocks – results can be found in tab.3
p.20. Finally, the binomial regression with (∆ × SOA − categorized) as predictors
(TsoaC) revealed the same Anticipation effect for Short SOA (β = −0.258 z =
−3.614 p < 0.0001) and Perceptual Delay for Unmasked conditions compared with
Long SOA (β = 0.58 z = 6.096 p < 0.0001) without any interaction. To sum up,
the effect of both metacontrast masking and SOA were comparable with the non-
motor blocks though with different size effects – see PSS in tab.3 p.20. Yet, a huge
difference between motor and non-motor conditions remains the fact the SOA and
∆ do interact in the motor case. At this stage, we could imagine two reasons for
this interaction:
bottleneck effect in case of positive ∆ (Sound after) and high SOA, the contiguity
D.R.L.Zarebski 21

3 Results
of Sound, Mask and Motor responses produce errors which are unrelated with
the timing of the motor response
motor interaction the interaction of SOA and ∆ is mediated by motor responses.
Quantification of the Motor effect in TOJ To investigate the reasons for this phe-
nomenon and quantify the effect of motor responses in a more precise way, we used
two different strategies.
Effect of motor responses We first modelled (TsoaMot) data of both motor and
non-motor blocks using, as we did before, participants’ intercepts as random factors.
The combination of three factors was used: ∆, SOA and a Motor variable –i,e. was
the data obtained with or without motor responses. Interestingly, we found an effect
of motor responses on the Prop(Target First) (β = −0.44 z = −2.92 p < 0.005)
which could be labelled as motor induced delay of the Target together with an
interaction of motor responses with ∆ (β = −0.0053 z = −2.46 p < 0.05) which
allowed us to conclude that the decreasing effect of motor responses on the proportion
of Target First gets stronger as ∆ reaches higher values. However, no interaction
was found between the Motor and SOA predictors (β = 0.0026 z = 1.85 p > 0.05).
Within trial analysis of Response times influences As highlighted by Cardoso-
Leite and Gorea 2010; Waszak and Gorea 2004, one of the main advantages of testing
TOJ and response times in the same block is the possibility to measure, trial per trial,
the effect of response times on Temporal Order Judgments. We thus modelled (Trt)
the proportion of Target First using the combination of (∆ × ResponseTimes) as
predictors. We thus found that not only are TOJ sensitive to the fact that a certain
Motor Response was given but also the fact Response Times influence the perception
of the order, for we found a negative effect of RT on Target First proportion (β =
−0.0025 z = −6.047 p < 0.0001) together with an interaction (β = −0.000026 z =
−5.015 p < 0.0001) of these two predictors. In other words, the slower the subject is,
the less likely Target will be perceived first. Adding SOA predictor to perform a new
analysis (Tsoart ∆ × SOA × RT) on masked trials only, we did not find any effect
of SOA (β = 0.00023 z = 0.105 p > 0.9) nor interaction with the other predictors
which is coherent with our replication of the Fehrer-Raab Effect.
summary of TOJ analysis To sum up, there is a general Anticipation effect in-
duced by metacontrast masking which is modulated by both motor responses and
SOA depending on the situation. In a non-motor context, the Anticipation effect is
compensated in long SOA where the proportion of Target First is similar as in Un-
masked conditions. However, in the motor context, the influence of RT together with
i) the replication of the Fehrer-Raab Effect, ii) the fact that we found interactions

3.4 Subjective Estimations (N=8)
600 800 1000 1200 1400
600800100012001400 no
150
0
120
90
60
30
600 800 1000 1200 1400
600800100012001400
no
120
30
150
60
0
90
Delay (ms)
EstimDelays(ms)
(a) by SOA
600 700 800 900 1100
600700800900100011001200
long
short
no
600 700 800 900 1100
600700800900100011001200
long
no
short
Delay (ms)
EstimDelays(ms) (b) SOA grouped
Figure 13: Subjective Estimations averaged across participants: x-axis represent physical delays
and y-axis their subjective estimations. Fig.13(a) details every SOA. Fig.13(b) groups conditions in
Long SOA –90, 120,150 ms– Short SOA –30,60ms– and Unmasked conditions. Details per subjects
can be found in fig.24 p.44.
between ∆ and SOA and ∆ and RT and iii) that fact that the effect of SOA vanishes
as soon as RT become predictor strongly suggest that the influence of SOA on the
Prop(Target First) is mediated, if not bypassed, by the speed of Motor Responses
– see.4.3 p.29 for a discussion of the implications of this finding.
3.4 Subjective Estimations (N=8)
For similar reasons as in TOJ analysis, we used Linear Mixed Models based on Re-
stricted Maximum Likelihood (REML) fits using participants’ intercepts as random
factors.
3.4.1 Non-motor blocks
Effects of Masking and SOA We started by modelling (Smask) the effect of mask-
ing vs control conditions (Unmasked and SOA0
) using Delay × Masking as pre-
dictors. Contrary to what happened in Temporal Order Judgments, metacontrast
masking increased the estimated time (β = 57.47 t = 3.88 p < 0.001) and seemed
to interact with the objective Delay between the fixation and the Target onset
(β = −0.052 t = −3.69 p < 0.001). In other words metacontrast masking seems to
delay the occurrence of the Target in a way which decreases for long Delays.
We then looked for a proportional effect of SOA using it in combination with
Delay as predictors on masked data only (SOA0
excluded). It turned out that the
estimation of the Delay was enhanced by SOA increasing (β = 0.5495 t = 2.054 p <
D.R.L.Zarebski 23

3 Results
0.05) without any interaction with the objective Delay (β = 0.00016 t = −0.651 p >
0.5) – see fig.13(a). Reformulated, this result means that, for the same objective time
between the Fixation and Target onsets, subjects tend to estimate the Target later if
the SOA is long in a way which do not depend on the occurrence of the Target in the
trial. Thus, while metacontrasted Targets are, in Subjective Estimations, delayed,
the effect induced by SOA on subjective temporality remained the same in Subjective
Estimations and Temporal Order Judgments.
Long vs Short dichotomy Applying the same subsetting method as for Temporal
Order Judgments –see.3.3.1 p.20 for details– with simple (Delay×SOA categorized)
models, we found that, among the 15 possible partitioning, (30,60) (90,120,150) was
better fitted than the other. In other words, the step in metacontrast masking func-
tions postulated in Reeves 1982; Sackur 2013 seems resistant to protocol change.
We thus modelled (SsoaC) the effect of conditions categorized in No masking,
Long and Short SOA (Delta × SOAcategorized) and found that, contrary to what
we found in Temporal Order Judgments, both Unmasked (β = −64.66 t = −3.50 p <
0.001) and Short SOA (β = −47.55 t = −2.03 p < 0.05) conditions were anticipated
compared with the Long SOA conditions –see fig.13(b). Interestingly the objective
Delay interacted in No mask conditions (β = 0.0486 t = 2.73 p < 0.01) but did not
for Short SOA (β = 0.0167 t = 0.74 p > 0.1).
To sum up, while producing, globally, a Perceptual Delay rather than an An-
ticipation effect, Subjective Estimations remains consistent with Temporal Order
Judgments regarding i) the delaying effect of SOA increasing ii) the existence of a
step in metacontrast masking function between 60 and 90 ms –see sec.4.2 p.28 for
a discussion of this difference.
3.4.2 Motor blocks
Replication of the analysis for Motor Blocks We first looked at the global masking
effect on Subjective Estimations by mean of a (Delay × Masking) combination
of predictors (Smask). We found that, while Motor conditions also produce the
Perceptual Delay we found in non-motor blocks, this effect was marginally significant
(β = 29.26 t = 1.97 p = 0.0484) as was the Delay × Masking interaction (β =
−0.0268 t = −1.88 p = 0.0599). Modelling (Ssoa) the effect of SOA combined
with the Delay as predictors, we did not find any effect of SOA (β = −0.161 t =
−0.602 p > 0.5) though we found a marginally significant effect of the Delay ×SOA
interaction (β = 0.0005 t = 2.03 p < 0.05) on Subjective Estimations.
For this very reason, modelling the effect of SOA categorized (SsoaC) on Subjec-
tive Estimations preceded by motor responses failed to see any difference between
Long SOA and short SOA (β = −25.98 t = −1.39 p > 0.1) nor No-mask conditions
(β = 9.18 t = 0.40 p > 0.5). Interactions with the Delay were not significant. In

other words, the Perceptual Delay induced by masking do not depends on SOA but
might depend on motor information as in the case of Temporal Order Judgments.
Quantification of the Motor effect in ESTIM To understand the effect of Motor
responses, we applied the same trial per trial analysis as we did for Temporal Order
Judgments. We first analysed the trial per trial effect of Response Times (Srt)
with (Delay × RT) as predictors. We found a positive effect of RT on Subjective
Estimations (β = 0.71 t = 6.240 p < 0.0001) together with an interaction with Delay
(β = −0.0007 t = −6.219 p < 0.0001) which show that the slower the subject is,
the later she would estimate the occurrence of the target though this effect decrease
slowly with the delay.
Surprisingly, by modelling (Ssoart) the interaction of (Delay × RT × SOA), we
did not find any effect of RT (β = 0.389 t = 1.447 p > 0.1) nor SOA (β = −1.117 t =
−1.397 p > 0.1) which could be interpreted as the fact that perceptual informations
related with SOA are altered by RT rather than merely bypassed.
summary of ESTIM analysis To sum up, beyond the general underestimation of
the delay, we found similar effects on perceived temporality in Subjective Estima-
tions and Temporal Order Judgments– see sec.4.2 p.28 for a discussion. In metacon-
trast masking, Target are estimated earlier (Anticipation effect) or later (Perceptual
Delay) than in the control conditions depending on the SOA. Increased SOA tends to
delay proportionally the subjective occurrence of the target which results, compared
with the control conditions, in a Anticipation effect for short SOA and a Perceptual
Delay for Long SOA. We also found a motor induced delay correlation of which has
been observed in-trial with response times – see sec.4.3 p.29 for a discussion.
4 Discussion
Results detailed in sec.3.4.2 put new perspectives on dual stream hypotheses. Before
discussing these implications –sec.4.3 p.29– we shall first focus our attention on
the Perceptual Delay – see sec.4.1 – and the complementarity of Temporal Order
Judgments and Subjective Estimations –see sec.4.2.
4.1 Anticipation effect vs Perceptual Delay
Interestingly, we did not reproduce the original Perceptual Delay of Didner and
Sperling 1980 but some kind of Anticipation effect. We propose three explanations
for this finding. First of all, it should be emphasized that the original experiment
of Didner and Sperling 1980 involved a later confidence judgment trial per trial
categories of which could have influence the results. Confidence levels were as follow:
D.R.L.Zarebski 25

4 Discussion
Model Predictors Data subset Interactions
SOA SOA cat RT Motor Masking
Tmask + + + no-MR ∅
Tmask + + + MR ∅
Tsoa − − − no-MR,mask ∅
Tsoa − − − MR, mask − − − (∆ × SOA)
TsoaC − − −(no) + + +(short) no-MR ∅
TsoaC − − −(no) + + +(short) MR ∅
TsoaMot − − − −− −(∆ × Motor)
Trt − − − MR − − −(∆ × RT)
Tsoart ∅ −− MR mask −−(∆ × RT)
Smask ++ no-MR −−(Delay × Masking)
Smask + MR ∅
Ssoa ++ no-MR, Mask ∅
Ssoa ∅ MR, Mask −(Delay × SOA)
SsoaC −−(no) −−(short) no-MR +(Delay × No)
SsoaC ∅ MR ∅
Srt ∅ + + + MR − − −(Delay × RT)
Ssoart ∅ ∅ MR, mask +(Delay × SOA)
Table 4: Summary of models used for analysis and their effects: T family of models (for Temporal
Order Judgments) were binomial regression using always ∆ as predictor. S family of models (for
Subjective Estimations) were linear regressions based on Restricted Maximum Likelihood ratio
using always the objective interval between fixation and target onsets (Delay) as a predictor. Data
subset indicates on which subset the model was performed. ∅ stands for no effect, RT for Response
Times, Motor for TOJ performed after a motor response. The number of + and - symbol indicate
the level of confidence.
certain
moderately certain
uncertain ”indicating that although the subject did not think the auditory stimulus
occurred at the same time as the visual activity onset, there was uncertainty as
to which occurred first” (Didner and Sperling 1980:237)
simultaneous
error the wrong temporal order button has been pressed
Figure 14: Sound vs Target First: sug-
gestion of the asymmetry of these two
conditions
Combining these levels and excluding errors
with Temporal Order Judgments, we thus have
7 categories from ”Target First surely” to ”Click
First surely”. Despite the fact that asking the
subjects to perform forced choice Temporal Or-
der Judgments with the possibility to report
negative –uncertainty– and positive simultaneity
may sound odd, there may be a problem with the
way these confidence judgements were used in the analysis. Because three of the four
subjects did not use the full range of these 7 categories, Didner and Sperling choose
to restrict the analysis to categories precision of which correspond to the degrees of
confidence of the most consistent categories across subject – i,e. conform both in

4.1 Anticipation effect vs Perceptual Delay
(a) round shape device (b) linear device (c) effect of device variation
Figure 15: Varieties of metacontrast masking stimuli
fig.15(c) displays the percentage of correct detection as a function of the delay of the mask de-
pending of the distance between the contour of the Target from the inner contour of the mask (in
degrees of visual angle)
From Lefton 1973
order and difficulty with the objective stimulus; click first with high and medium
certainty.
This suppression of noise before analysis based on trans-subjective criterion could
have been acceptable if and only if the task were a symmetric one. However, given
the fact that, in Target First condition, the contiguity of Sound and Mask could
produce noise –see fig.14– we could imagine that, in such a noisy situation, the
subject would be less confident about his Temporal Order Judgments. In other words
a more important portion of his Target First answers would have been uncertain
thus discarded from analysis producing a higher proportion of Click First answers
on positive ∆ conditions. Though Didner and Sperling 1980 is not detailed enough
to support this explanation, the original Perceptual Delay could well be an artefact
of this second order task.
A second reason might be related with the stimulus configuration. As stated
before, it is possible to produce metacontrast masking with different stimuli from
the linear stimulus of Didner and Sperling 1980 –see fig.15(b)– to our own square
shapes –see fig.4 p.8. Yet it has been acknowledged that the variety of these devices
do not produce the same amount of lateral inhibition thus masking (Lefton 1973 see
fig.15(c)). Given that the time perception in metacontrast masking seems closely
related with visibility – see sec.4.1.1 p.28 for details– we could imagine that the
variation of the masking function across studies might lead to different biases.
Thirdly, though related with the previous point, we could imagine that the type
A / type B dichotomy of metacontrast masking functions (Kolers 1962) might
be involved. While type B metacontrast masking, which occurs when the Target
is more energetic than the Mask as in our experiments, is known to produce the
common U-shaped function of visibility across SOA, visibility function in type A
metacontrast masking, which occurs when the Mask is more energetic than the
D.R.L.Zarebski 27

4 Discussion
Target as in Didner and Sperling 1980 experiment, is a linear one. Given that there
might be, at least in non-motor context –see sec.4.3 p.29– a subtle link between
visibility and time perception in Temporal Order Judgments, one could imagine that
type A and type B metacontrast masking may produce different Temporal Order
Judgments functions.
4.1.1 Long, Short SOA and the dimensions of visibility in metacontrast masking
The clear dissociation between long and short SOA has also some equivalent in the
literature of brief events perception. We could first cite the multimodal Judgement
of simultaneity experiment of Efron 1970 whose results suggest that 120ms is the
subjective threshold of discriminability of events’ durations. In other words, no
matter how variable, every simple event would be perceived as a 120ms long event.
More directly correlated, (Sackur 2013)’s analysis of visibility in metacontrast
masking by mean of a multidimensional scaling method – i,e. the creation of a
distance matrix based on judgements of similarity – suggests that, while long and
short SOA could lead to equivalent visibility judgements –thus the a U phenomenon–
similarity judgements lead to the conclusion that the very notion of visibility, too
inclusive, involves different mechanisms for long and short SOA. While short SOA
may lead to the perception of one item through some form of integrative processes
(Cass and Alais 2006; Eriksen and Collins 1967), long SOA may not involve such an
integration for physiological reasons such as the lack of persistence in iconic memory
(Coltheart 1980; Coltheart and Arthur 1972; Di Lollo 1980; Hogben and Di Lollo
1974).
Moreover, these integration of spatial features trough time are known to inter-
vene on high level processing such as subjective rating of the degree of temporal
integration, partial report performance tasks (Loftus and Irwin 1998) and visual
word recognition (Forget, Buiatti, and Dehaene 2010) under a certain critical SOA’s
duration (≃ 80ms). Thus, the existence of a discontinuity in SOA’s impact on
perceived temporality together with the consistency of such a coarse grained tem-
porality across our protocol of measure –see sec.4.2 for a detailed discussion– might
be related with the competition of a integrative processes with a dissociative one in
a similar fashion as Reeves 1982 –see sec.1.3.1 for details.
4.2 TOJ and ESTIM consistency
One of the most striking result is probably the complementarity of Temporal Order
Judgments and Subjective Estimations. Though the possibility of Subjective Es-
timations has been shown for experimental durations such as SOAs (Marti et al.
2010), the fact that introspection can reliably access durations ten times higher than
those intervening in Temporal Order Judgments is even more surprising given the

4.3 Motor interactions
influence of attentional variations on time perception (Enns, Brehaut, and Shore
1999). In a similar way, given the unimodality of Subjective Estimations, one might
have expected that spatial attention enhanced the temporal resolution as in the two
flashes segmentation paradigm of Yeshurun and Levy 2003.
Finally, given the great number of adaptation mechanisms for the perception of
simultaneity in multi-modal contexts – attentional asynchrony effect (Sinnett et al.
2007) such as the multisensory prior entry effect (Spence, Shore, and Klein 2001;
Spence and Squire 2003), recalibration to compensate for sound lags (Fujisaki and
Nishida 2005; Vroomen et al. 2004) and modality specific timing (Bueti, Bahrami,
and Walsh 2008; Hirsh and Sherrick 1961) – the consistency of Temporal Order
Judgments and Subjective Estimations is not a trivial result.
Consistent, not identical Nevertheless, some differences between Temporal Order
Judgments and Subjective Estimations remain to be explained. Though consistent
with each other, the degree of delay induced by metacontrast masking is not the
same. While metacontrasted Targets are overall anticipated in Temporal Order
Judgments, Subjective Estimations seems to delay Targets’ occurrence. The fact
that SOA nonetheless delays the target proportionally despite the general effect
in metacontrast masking suggest that both tasks relies on a common information
modulated by the nature of the task.
Figure 16: Exponential effect in Sub-
jective Estimations: estimation (E) of
the magnitude of a given physical value
(SREL)
from Borg 1990
Among its well documented effects, it should
be emphasized that Subjective Estimations are
known to behave exponentially. As an exam-
ple, if one is asked to reduce her speed (say 100
Km/h) until she perceives it to be half fast, the
objective speed (S2) would be 70 Km/h rather
that 50 Km/h – see fig.16 and Borg 1990. How-
ever, this particularity does not explain easily
the overall Perceptual Delay induced by Subjec-
tive Estimations, for control non-masked condi-
tions should have been equally delayed. We spec-
ulate that the overall Perceptual Delay might
be related with some over-intellectualisation in-
duced by the multi-valuation of this protocol though we lack a precise explanation.
Summarized results We found that, though SOA did not impact on Response Times
(Fehrer-Raab Effect– see sec.3.2), Response times interact nonetheless with percep-
tual tasks – Temporal Order Judgments (sec.3.3.2 p.21) and Subjective Estima-
tions (sec.3.4.2 p.24) – which suggest that these tasks are mediated – either altered
D.R.L.Zarebski 29

4 Discussion
(a) (b)
Figure 17: The underlying logic of the Anticipation Response Times paradigm: stimuli of different
salience –see fig.17(a)– produces different internal latencies – see fig.17(b). The perceptual delay
induced by the black stimulus compared with the grey one is the difference between their respective
responses (∆ART )
From Cardoso-Leite and Gorea 2010
or by-passed – by motor responses. In general, temporal modulations (Anticipation
effect and Perceptual Delay) were less salient when performed with motor tasks.
Figure 18: Variations of perceptual and
motor latencies: meta-analysis of four
different studies
from Cardoso-Leite and Gorea 2010
One could argue that such an interaction is
solely the by-product of some bottleneck effect,
that the very fact of performing a motor task
before Temporal Order Judgments or Subjective
Estimations biases these later components of the
task. However, the observation of co-variations
of RT and perceptual task trial per trial excludes
this block-based explanation of the results.
We shall first detail some of the histori-
cal models for path dissociations (resp. inter-
actions), their experimental methodologies and
predictions – see sec.4.3.1 – before discussing
the empirical and methodological implications of
our results on this field – see sec.4.3.2. Finally
we will propose further complementary investi-
gations susceptible to answer some of the unsolved issues raised by our study –
sec.4.3.3 p.36.
4.3.1 Late developments on perceptual and motor latencies
As we pointed it out in the Introduction, one of the most direct approach on dorso
ventral interaction consists in choosing an experimental variable and measure its
impact on
perceptual latencies Temporal Order Judgments or, more recently, the Anticipa-
tion Response Times (ART) paradigm (Cardoso-Leite, Mamassian, and Gorea
2009) see fig.17

motor latencies Simple Response Times
Perceptual and motor latencies are known to vary across a great number of con-
ditions. However, the issues of whether they i) covariate ii) in a proportional way
are more subtle ones. Though a great number of studies suggests that RT-PSS
co-variance is resistant to the variation of a great number of parameters such as
salience (Adams and Mamassian 2004), spatial frequency (Barr 1983; Tappe, Nie-
pel, and Neumann 1994), luminance rise times (Ja´skowski 1993) or stimulus duration
(Ja´skowski 1991, 1992), the proportion of delay induced by experimental condition
in, respectively, motor and perceptual systems is not the same. The motor part is
known to be more delayed than the perceptual one (see. fig.18).
Three models for RT TOJ dissociations According to Cardoso-Leite and Gorea 2010,
explanatory models of simple RT-TOJ dissociation can be classified as
Two Independent Pathways (2IP) models which suggest that RT-TOJ dissoci-
ations result from the clear dissociation of different cerebral structures (Neu-
mann, Esselmann, and Klotz 1993; Tappe, Niepel, and Neumann 1994)
Single Pathway, Single Decision (1P1D) models which posit the same RT and
TOJ subtending processes predicting thus that every experimental variable
would affect equally those two tasks (Gibbon and Rutschmann 1969) and
One pathway, Two Decisions (1P2D) models explanations of which lie in the
idea that, though dependent on the same internal response to the stimulus,
RT TOJ dissociations reflect thresholds or criterion differences (Sternberg and
Knoll 1973)
Because of their strong predictions, 2IP are vulnerable to within trial analysis
of RT and PSS variance such as Cardoso-Leite, Mamassian, and Gorea 2007’s or
our own. As we suggested it earlier, the case of 1P1D is a more subtle one, for
1P1D’s predictions do not concern RT and PSS co-variance but the proportional
co-affectation of both measures. However, many studies have shown a greater af-
fectation of motor lags induced by luminance variations (Ja´skowski 1992), attention
or sensory channel selection (Jaskowski and Verleger 2000), contrast with back-
ground luminance (Menendez and Lit 1983; Roufs 1974) –see fig.18. Different ad
hoc hypotheses have been proposed. One of the most crucial rests on the fact that
RT and TOJ were most of the time blocked thus tested separately; driving to the
parsimonious explanation that the unequal affectation of TOJ and RT might be re-
lated with some higher attentional effect over the first than the second. Again, this
type of explanation does not hold against within trial RT TOJ combination based
experiments.
D.R.L.Zarebski 31

4 Discussion
Figure 19: Two criteria model for RT
TOJ dissociation
from Cardoso-Leite and Gorea 2010
Finally, 1P2D explain the difference in mag-
nitude of perceptual and motor latencies – see.18
– suggesting that the motor criterion might be
higher than the perceptual one thus more af-
fected (Cardoso-Leite, Mamassian, and Gorea
2007; Miller and Schwarz 2006; Sanford 1974)
– see fig.19 for an illustration of the phenomenal
(resp. motor) delays induced by a low energetic
(El) and a high energetic (Eh) stimulus. One of
the most central assumption for some of these
models is the idea that i) the internal response
increases linearly in time and that ii) the slope is proportional with the stimulus
intensity (Carpenter 1981; Ejima and Ohtani 1987). In a similar way, models such
as those proposed by Waszak, Cardoso-Leite, and Gorea 2007; Waszak and Gorea
2004 suggest that the motor criterion might be fixed while the perceptual one could
vary depending on the nature of stimuli and tasks.
Main limitation: predictive dichotomy of the effect of one experimental variable To
sum up, besides the subtle issue of the magnitude of perceptual and motor latencies,
both experimental procedures and predictions of these different families of models are
straightforward. The typical prediction of the Dissociation View states that there
might be some conditions in which one system would remained unaltered while the
One Process View predicts that every latencies of one system (or decision) comes
with latencies on the other.
However, we would like to suggest, in the next section, that such an experimental
dichotomy remains limited because i) it relies on a non contrastive approach of
purely perceptual vs motor-perceptual contexts – either testing response times and
Temporal Order Judgments in different blocks or both in the same trials – ii) and
consists in choosing an experimental variable to see whether it affect both paths or
not.
4.3.2 Two interacting systems
We would like to suggest that the main question is not to know whether a certain
experimental variable would affect both perceptual and motor tasks nor whether
it would affect these tasks equally but, rather, to know in which context one in-
formation is susceptible to be used in both tasks. We will first solve the apparent
contradictory replication of the Fehrer-Raab Effect.
Motor interaction and Fehrer-Raab Effect The fact that we obtained an effect of
response time on time perception though we replicated the Fehrer-Raab Effect is an

(a) in Temporal Order Judgments (b) in Subjective Estimations
Figure 20: Causal representation of motor/perceptual interactions: visual summary of our results
for Temporal Order Judgments (fig.20(a)) and Subjective Estimations (fig.20(b)). Solid lines
represent significant effects or interactions. Dashes lines represent marginally significant effects or
interactions. See also table 4 p.26 for a summary.
interesting kind of asymmetry we would like to focus on. We found that, while we
could replicate the Fehrer-Raab Effect – see sec.3.2 p.16 and tab.3 p.20 – Temporal
Order Judgments and response times interact trial per trial; suggesting that the
motor and phenomenal latencies co-variate – see sec.3.3.2 p.21 and see sec.3.4.2
p.24. This finding could sound self contradictory if one forget that these two finding
do not intervene at the same level.
Indeed, the Fehrer-Raab Effect has often been interpreted as an experimental
evidence for the Dissociation View (Neumann and Klotz 1994) – see Cardoso-Leite
and Gorea 2010:110-112 for a review– until Waszak and Gorea 2004 whose detection
variant of Fehrer and Raab 1962 –go task followed by a detection task in the same
trial– suggested that response times decreased with Targets intensity only when they
were visible. However, besides the question of the visibility of the Target, 7
a crucial
point should be highlighted.
A parsimonious interpretation of the Fehrer-Raab Effect as our does not rely on
the notion of visibility nor on any tacit relation between visibility and perceptual
latencies. The most pedestrian interpretation for the Fehrer-Raab Effect states
that response times do not vary according to SOA which does not exclude other
source of variability such as attentional resources, internal connectivity or, relative
to protocols, experimental variables such as ∆ or delay – see fig.20 for a causal
representation of these interactions.
To sum up, the replication of the Fehrer-Raab Effect entails a 2IP only if one
either focus on the co-variance of perceptual and motor latencies relative to one
experimental parameter (here, SOA) or ii) reduce perceived temporality to visibility.
An asymmetrical effect Indeed, have response times been also affected, we could
have proposed an explanation similar to Milner and Dyde 2003 or Steglich and
7As stated earlier, Targets were visible in every conditions, though we did not specifically test the visibility nor
other phenomenal characteristics.
D.R.L.Zarebski 33

4 Discussion
(a) Dissociation View (b) One Process View (c) Interacting systems
Figure 21: Explanations for RT-PSS co variations: fig.21(a) represents the possible explanation of
2IP models: the same information is sent in both systems. Fig.21(b) represents the explanation of
either 1P1D or 1P2D models: one system responsible for both tasks. Finally, fig.21(c) stands for
the two interacting systems we suggest.
Neumann 2000 which, roughly formulated, states that a similar low level and non-
contextual information is sent in both motor et perceptual systems – see fig.21(a)
and the Introduction p.6 for details. In a similar fashion, a supporter of the One
Process View could also, with the very same data, argue that the fact that both
tasks seems impacted by the same experimental variable involve that there is only
one system responsible for perceptual and motor tasks – see fig.21(b). However, we
hope to have shown that this alternative is not a decidable one, for most studies did
not test Temporal Order Judgments and Response times both blocked and trial per
trial the way we did.
We might propose the following explanation for this asymmetrical effect of SOA
on the perceived onset of Targets.
Two systems The fact that the same low level informations (SOA and spacial
frequencies responsible for metacontrast masking) do not impact both tasks
independently suggests that these tasks relies on different systems sensitive to
different informations.
interact However, this does not mean that these systems are independent as 2IP
suggested it, for we know that perceptual tasks (both Temporal Order Judg-
ments and Subjective Estimations) are influenced by motor informations (SOA
is not a necessary and sufficient information in this context).
depending on the context This use of motor information seems dependant on
the context – i,e. whether or not there is a relevant motor information. For
this very reason, testing perceptual and motor effects in different blocks or trial
per trial has a crucial impact on the informations used by the perceptual task.
We could speculate, given the different sensitivities of these putative systems,
that the interaction in motor context occurs after some informational split between
informations with and informations without SOA related properties – see fig.21(c)

for an illustration. Finally, given the speed differences between ventral and dorsal
regions of the visual cortex, if these paths actually implement the two systems we
postulated, their interaction in motor context would perfectly make sense from an
evolutionary point of view: from two concurrent informations, the more reliable one
is used.
Is the perceptual task still perceptual? One could argue that these motor influ-
ences on perceptual tasks do not mean that the perceptual information is altered
but, rather, merely by-passed. Closely related with the conscious non-conscious dis-
tinction often superimposed on path dissociation, this intuition implies that Tem-
poral Order Judgments and Subjective Estimations might be guided by merely
unconscious motor informations which contradict the unchanged ”real phenomenal
temporality”.
There is two different levels to answer this question. The first one is method-
ological. Tasks such as Temporal Order Judgments have been confidently used as
behavioral manifestation on one’s subjective experience for decades. Doubts may
be raised about some of these operational criterion of one’s conscious experience.
However, it should be kept in mind that there is no way to know what the subject
experienced but to ask him. Yet, the same dissociative argument between one’s an-
swers and one’s conscious experience might be raised for any other perceptual tasks
we could imagine.
The second level, related with this methodological issue, is a conceptual one.
The idea that conscious phenomenal experience takes place in ventral regions of
the visual cortex while frontal regions implement some kind of access consciousness
(Block 1998, 2005) is committed in local conception of phenomenal awareness such as
the Recurrence Based Theory (Lamme 2004; van Gaal and Lamme 2012). From the
point of view of these theories, it is perfectly possible to claim that what is accessible
and expressed in perceptual tasks such as Temporal Order Judgments and Subjective
Estimations – i.e. mixed phenomenal and motor informations – differs from pure
phenomenal informations.
However, this theoretical commitment remains only one option among many al-
ternative Theories of Consciousness – see de Gardelle and Kouider 2009 for a re-
view. As an example, the Global Workspace Model (Dehaene and Changeux 2011;
Kouider, Sackur, and de Gardelle 2012), i) do not distinguish phenomenal from ac-
cess consciousness ii) suggesting that consciousness is an holistic mental activity
which involves frontal regions (Del Cul et al. 2009). From this point of view, the
fact that an information, though sufficient for perceptual consciousness in non-motor
context, could be mediated by motor-based informations does not threat the gen-
uine perceptual nature of what is reported 8
. This issue, experimentally undecidable
8We could also mention Sensorimotor approaches of phenomenal consciousness (O’Regan 2009; O’Regan and
D.R.L.Zarebski 35

4 Discussion
from the perspective of the present study, remains an open one.
4.3.3 Possible refinements
Among the limitations of the present study, the fact that we did not test the vari-
ation of visibility trial per trial together with perceived time and motor responses
hindered our discussion of Waszak and Gorea 2004’s alternative explanation of the
Fehrer-Raab Effect –i.e. decreasing RTs depends on target visibility. Given the
central place of this argument for the One Process View and the curious, though
uncertain, relation entertained by visibility in type B metacontrast masking and
time perception – i.e. the fact that the step we found for Perceptual Delay (see
sec.3.3.1 p.20) coincides with what is known as the minimal visibility for SOA around
80ms (see sec.4.1.1 p.28) – testing visibility might allow for a better understanding
of these dependencies.
An other way to distinguish these systems rigorously might be to find a variable
which would only affect Response time realized alone though we could predict that,
in a similar way as SOA in our experiment, its effect might vanish in a mixed
perceptual motor context. The issue remains to find such a variable, for most of the
low level variables which affect Response times are known to also affect Temporal
Order Judgments – see sec.4.3.1 p.30.
Finally, we only instantiated motor and perceptual systems from i) their tasks and
ii) the type of informations used. However, the possibility that these systems might
actually be implemented by ventral and dorsal paths of the visual cortex remains
an open question. Yet, the complementarity of Temporal Order Judgments and
Subjective Estimations – see sec.4.2 p.28 – is a new step forward for a possible neural
objectivation of these mechanisms inasmuch as, Perceptual Delay being detectable
trial per trial in the latter case, electroencephalography based investigations might
be subsequently facilitated in Subjective Estimations.
Conclusion
By distinguishing blocked and trial per trial perceptual and motor tasks combina-
tions, we found that we could replicate the Fehrer-Raab Effect and the modulations
of perceived temporality (Perceptual Delay or Anticipation effect) by SOA in meta-
contrast masking. This finding contradicted the classical One Process View. Yet,
revelling a mediation of Response times on both perceptual tasks (Temporal Order
Judgments and Subjective Estimations), our results did not support the Dissociation
View either.
Noë 2001) which reject the vision for perception vision for action distinction in a similar way as Cardoso-Leite and
Gorea 2010; Cardoso-Leite, Mamassian, and Gorea 2009.

References
We then suggested that i) there might be two different systems sensitive to dif-
ferent kinds of low level visual features ii) able to interact in dual perceptuo-motor
context. The conclusion is thus both empirical and methodological, for the fact that
perceptual tasks might use different informations depending on the context – SOA
vs mixed RT-SOA– could only been highlighted by our contrastive method which
has not been, to our knowledge, used in dual stream investigations.
Yet, this finding raised a knew issue. Though it is parsimonious to conclude
that Temporal Order Judgments cannot be used in perceptuo-motor context as the
exact counterparts of pure perceptual context, the issue of whether these perceptual
tasks remain perceptual – i.e. whether they still reflect the subjective perceptual
experience rather than purely unconscious motor informations – remains an open
question because of its ramifications with debates about the nature of consciousness.
References
Adams, W. and P. Mamassian (2004). “The effects of task and saliency on latencies for colour
and motion processing”. In: Proceedings of the Royal Society of London. Series B: Biological
Sciences 271.1535, 139–146.
Albrecht, T., S. Klapötke, and U. Mattler (2010). “Individual differences in metacontrast masking
are enhanced by perceptual learning”. In: Consciousness and cognition 19.2, 656–666.
Allan, L., A.B. Kristofferson, and E. W. Wiens (1971). “Duration discrimination of brief light
flashes”. In: Perception & Psychophysics 9.3, 327–334.
Allan, L.G. and A.B. Kristofferson (1974). “Judgments about the duration of brief stimuli”. In:
Perception and Psychophysics 15.3, pp. 434–440.
Bachmann, T. (2010). “Individual differences in metacontrast: An impetus for clearly specified
new research objectives in studying masking and perceptual awareness?” In: Consciousness
and cognition 19.2, 667–671.
Barr, M. (1983). “A comparison of reaction-time and temporal-order judgment estimates of latency
to sinusoidal gratings”. In: Perception 12.7.
Bertelson, P. and G. Aschersleben (2003). “Temporal ventriloquism: crossmodal interaction on the
time dimension: 1. Evidence from auditory–visual temporal order judgment”. In: International
Journal of Psychophysiology 50.1–2, pp. 147–155. issn: 0167-8760.
Bevan, W., J. Jonides, and S. C. Collyer (1970). “Chromatic relationships in metacontrast sup-
pression”. In: Psychonomic Science 19, 367–368.
Block, N. (1998). “On a confusion about a Function of Consciousness”. In: The Nature of Con-
sciousness. N.Block, O.Flanagan, G.Güzeldere (eds). MIT Press, pp. 375–416.
— (2005). “Two neural correlates of consciousness”. In: Trends in cognitive sciences 9.2, 46–52.
Borg, G. (1990). “Psychophysical scaling with applications in physical work and the perception of
exertion”. In: Scandinavian journal of work, environment & health, 55–58.
Breitmeyer, B.G. and L. Ganz (1976). “Implications of sustained and transient channels for theories
of visual pattern masking, saccadic suppression, and information processing.” In: Psychological
review 83.1, 1–36.
Breitmeyer, B.G. and H. Ogmen (2006). Visual Masking: Time Slices Through Conscious and
Unconscious Vision. en. Oxford University Press. isbn: 9780198530671.
Breitmeyer, B.G. et al. (2006). “Meta-and paracontrast reveal differences between contour-and
brightness-processing mechanisms”. In: Vision Research 46.17, 2645–2658.
Bueti, D., B. Bahrami, and V. Walsh (2008). “Sensory and association cortex in time perception”.
In: Journal of Cognitive Neuroscience 20.6, 1054–1062.
D.R.L.Zarebski 37

References
Cardoso-Leite, P. and A. Gorea (2010). “On the perceptual/motor dissociation: a review of con-
cepts, theory, experimental paradigms and data interpretations”. In: Seeing and Perceiving 23.2,
89–151.
Cardoso-Leite, P., P. Mamassian, and A. Gorea (2007). “Temporal order judgment and simple
reaction times: Evidence for a common processing system”. In: Journal of Vision 7.6.
— (2009). “Comparison of perceptual and motor latencies via anticipatory and reactive response
times”. In: Attention, Perception, & Psychophysics 71.1, 82–94.
Carpenter, R. H. S. (1981). “Oculomotor procrastination”. In: Eye movements: Cognition and visual
perception. Ed. by D.F. Fisher, R.A. Monty, and J.W. Senders. Lawrence Erlbaum Associates.
Hillsdale, NJ, 237–246.
Cass, J. and D. Alais (2006). “Evidence for two interacting temporal channels in human visual
processing”. In: Vision research 46.18, 2859–2868.
Coltheart, M. (1980). “The persistences of vision”. In: Philosophical Transactions of the Royal
Society of London. B, Biological Sciences 290.1038, 57–69.
Coltheart, M. and B. Arthur (1972). “Evidence for an integration theory of visual masking”. In:
The Quarterly Journal of Experimental Psychology 24.3, 262–269.
Corallo, G. et al. (2008). “Limits on Introspection: Distorted Subjective Time During the Dual-Task
Bottleneck”. In: Psychological Science 19.11, pp. 1110–1117.
Corveleyn, X., J. Lopez-Moliner, and Y. Coello (2012). “Motor action reduces temporal asynchrony
between perceived visual changes”. In: Journal of Vision 12.11, pp. 1–16.
Davison, A.C. (2003). Statistical models. Cambridge Series in Statistical and Probabilistic Mathe-
matics 11. Cambridge University Press.
De Gardelle, V. and S. Kouider (2009). “Cognitive Theories of Consciousness”. In: Encyclopedia of
Consciousness. Ed. by W.P. Banks. Vol. 1. Oxford: Elsevier, pp. 135–146.
Dehaene, S. and J. P. Changeux (2011). “Experimental and theoretical approaches to conscious
processing”. In: Neuron 70.2, 200–227.
Del Cul, A. et al. (2009). “Causal role of prefrontal cortex in the threshold for access to conscious-
ness”. In: Brain 132.9, 2531–2540.
Di Lollo, V. (1980). “Temporal integration in visual memory.” In: Journal of Experimental Psy-
chology: General 109.1, p. 75.
Di Lorenzo, J. and I. Rock (1982). “The rod-and-frame effect as a function of the righting of
the frame”. In: Journal of experimental psychology: human perception and performance 8.4,
536–546.
Didner, R. and G. Sperling (1980). “Perceptual delay: a consequence of metacontrast and apparent
motion.” In: Journal of Experimental Psychology: Human Perception and Performance 6.2,
pp. 235–243.
Dyde, R. and A.D. Milner (2002). “Two illusions of perceived orientation: one fools all of the
people some of the time; the other fools all of the people all of the time”. In: Experimental
Brain Research 144.4, 518–527.
Efron, R. (1970). “The minimum duration of a perception”. In: Neuropsychologia 8.1, 57–63.
Ejima, Y. and Y. Ohtani (1987). “Simple reaction time to sinusoidal grating and perceptual in-
tegration time: contributions of perceptual and response processes”. In: Vision Research 27.2,
269–276.
Enns, J.T., J. Brehaut, and D.I. Shore (1999). “The duration of a brief event in the mind’s eye”.
In: The Journal of general psychology 126.4, 355–372.
Eriksen, C. and J. Collins (1967). “Some temporal characteristics of visual pattern perception.” In:
Journal of Experimental Psychology 74.4p1, p. 476.
Exner, S. (1868). “Über die zu einer Gesichtswahrnehmung nöthige Zeit”. In: Wiener Sitzungsber
Math-Naturwissensch Cl Kaiser Akad Wissensch 58.2, pp. 601–632.
Faraggi, D., P. Izikson, and B. Reiser (2003). “Confidence intervals for the 50 per cent response
dose”. In: Statistics in medicine 22.12, 1977–1988.
Fehrer, E. (1965). “Contribution of the perceptual segregation to the relationship between stimulus
similarity and backward masking”. In: Perceptual and Motor Skills 21.1, 27–33.
— (1966). “Effect of stimulus similarity on retroactive masking.” In: Journal of Experimental
Psychology 71.4, p. 612.

References
Fehrer, E. and I. Biederman (1962). “A comparison of reaction time and verbal report in the
detection of masked stimuli.” In: Journal of Experimental Psychology 64, p. 126.
Fehrer, E. and D. Raab (1962). “Reaction time to stimuli masked by metacontrast.” In: Journal
of experimental psychology 63.2, pp. 143–147.
Fink, M. et al. (2006). “Stimulus-dependent processing of temporal order”. In: Behavioural processes
71.2, 344–352.
Finney, D.J. (1971). Probit Analysis. 3rd ed. New York: Oxford University Press.
Forget, J., M. Buiatti, and S. Dehaene (2010). “Temporal integration in visual word recognition”.
In: Journal of cognitive neuroscience 22.5, 1054–1068.
Franz, V. et al. (2000). “Grasping visual illusions: No evidence for a dissociation between perception
and action”. In: Psychological Science 11.1, 20–25.
Franz, V. H., F. Scharnowski, and K. R. Gegenfurtner (2005). “Illusion effects on grasping are
temporally constant not dynamic”. In: Journal of Experimental Psychology-Human Perception
and Performance 31.6, 1359–1377.
Fujisaki, W. and S. Nishida (2005). “Temporal frequency characteristics of synchrony asynchrony
discrimination of audio-visual signals”. In: Experimental Brain Research 166.3-4, 455–464.
Gibbon, J. and R. Rutschmann (1969). “Temporal order judgment and reaction time.” In: Science
165, pp. 413–415.
Goodale, M. and A.D. Milner (1992). “Separate visual pathways for perception and action”. In:
Trends in neurosciences 15.1, 20–25.
Goodale, M. and D. Westwood (2004). “An evolving view of duplex vision: separate but interacting
cortical pathways for perception and action”. In: Current opinion in neurobiology 14.2, 203–211.
Gorea, A., P. Mamassian, and P. Cardoso-Leite (2010). “Introspective duration estimation of re-
active and proactive motor responses”. In: Acta psychologica 134.2, 142–153.
Harrison, K. and R. Fox (1966). “Replication of reaction time to stimuli masked by metacontrast.”
In: Journal of Experimental Psychology 71.1, p. 162.
Heath, M., C. Rival, and G. Binsted (2004). “Can the motor system resolve a premovement bias
in grip aperture? Online analysis of grasping the Müller-Lyer illusion”. In: Experimental Brain
Research 158.3, 378–384.
Hirsh, I.J. and C.E. Sherrick (1961). “Perceived order in different sense modalities.” In: Journal of
experimental psychology 62.5, p. 423.
Hogben, J. and V. Di Lollo (1974). “Perceptual integration and perceptual segregation of brief
visual stimuli”. In: Vision Research 14.11, 1059–1069.
Jaskowski, P. and R. Verleger (2000). “Attentional bias toward low-intensity stimuli: an explana-
tion for the intensity dissociation between reaction time and temporal order judgment?” In:
Consciousness and cognition 9.3, 435–456.
Ja´skowski, P. (1991). “Perceived onset simultaneity of stimuli with unequal durations”. In: Percep-
tion 20, pp. 715–726.
— (1992). “Temporal-order judgment and reaction time for short and long stimuli”. In: Psycho-
logical research 54.3, 141–145.
— (1993). “Temporal-order judgment and reaction time to stimuli of different rise times”. In:
Perception 22, pp. 963–970.
Kahneman, D. (1967). “An onset-onset law for one case of apparent motion and metacontrast”. In:
Perception and Psychophysics 2, pp. 557–583.
— (1968). “Methods, findings, and theory in the studies of visual masking”. In: Psychological
Bulletin 70, pp. 404–425.
Keetels, M. and J. Vroomen (2005). “The role of spatial disparity and hemifields in audio-visual
temporal order judgments”. In: Experimental Brain Research 167.4, 635–640.
Klotz, W. and P. Wolff (1995). “The effect of a masked stimulus on the response to the masking
stimulus”. In: Psychological research 58.2, 92–101.
Kolers, P.A. (1962). “Intensity and contour effects in visual masking”. In: Vision Research 2.9,
277–IN4.
Kouider, S., J. Sackur, and V. de Gardelle (2012). “Do we still need phenomenal consciousness?
Comment on Block”. In: Trends in cognitive sciences 16.3, p. 140.
Lamme, V. A. F. (2004). “Separate neural definitions of visual consciousness and visual attention;
a case for phenomenal awareness”. In: Neural Networks 17.5-6, 861–872.
D.R.L.Zarebski 39

References
Lefton, L. (1973). “Spatial factors in metacontrast”. In: Perception & Psychophysics 14.3, 497–500.
Lewis, J.H., H. Matteson, and W. Dunlap (1977). “Effects of stimulus duration on perceptual
latency in the fovea.” In: Bulletin of the Psychonomic Society.
Loftus, G. and D.E. Irwin (1998). “On the relations among different measures of visible and infor-
mational persistence”. In: Cognitive psychology 35.2, 135–199.
Marti, S. et al. (2010). “Mapping introspection’s blind spot: Reconstruction of dual-task phe-
nomenology using quantified introspection”. In: Cognition 115.2, pp. 303–313.
Matteson, G. and T. Flaherty (1976). “A latency measure of metacontrast”. In: Bulletin of the
Psychonomic Society 7, pp. 545–547.
McDougall, W. (1904). “The sensations excited by a single momentary stimulation of the eye”. In:
British Journal of Psychology 1, pp. 78–113.
McIntosh, R. and T. Schenk (2009). “Two visual streams for perception and action: current trends.”
In: Neuropsychologia 47.6, p. 1391.
Menendez, A. and A. Lit (1983). “Effects of test-flash and steady background luminance on simple
visual reaction time (RT) and perceived simultaneity (PS)”. In: Investigative Ophthalmology
and Visual Science 24.5.
Miller, J. and W. Schwarz (2006). “Dissociations between reaction times and temporal order judg-
ments: a diffusion model approach.” In: Journal of experimental psychology: human perception
and performance 32.2, p. 394.
Milner, D. and R. Dyde (2003). “Why do some perceptual illusions affect visually guided action,
when others don’t?” In: Trends in Cognitive Sciences 7.1, 10–11.
Minkowski, M. (1911). “Zur Physiologie der Cortikalen Sehsphäre”. In: Pflügers Archiv European
Journal of Physiology 141.4, 171–327.
Mishkin, M. and L.G. Ungerleider (1982). “Contribution of striate inputs to the visuospatial func-
tions of parieto-preoccipital cortex in monkeys”. In: Behavioural brain research 6.1, 57–77.
Moscatelli, A., M. Mezzetti, and F. Lacquaniti (2012). “Modeling psychophysical data at the
population-level: The generalized linear mixed model”. In: Journal of vision 12.11, pp. 1–17.
Neumann, O. (1979). “Visuelle Aufmerksamkeit und der Mechanismus des Metakontrasts”. In:
Kongress der Deutschen Gesellschaft fur Psychologi. Ed. by L. Eckensberger. Gottingen: Hogrefe.
Neumann, O., U. Esselmann, and W. Klotz (1993). “Differential effects of visual-spatial attention
on response latency and temporal-order judgment”. In: Psychological research 56.1, 26–34.
Neumann, O. and W. Klotz (1994). “Motor responses to non-reportable, masked stimuli: Where
is the limit of direct parameter specification?” In: Attention and Performance XV: Conscious
and Non- conscious Information Processing. Ed. by C. Umiltà and M. Moskovitch. MIT Press.
Cambridge, MA.
Neumann, O. et al. (1992). “Reaction time and temporal serial judgment: Corroboration or disso-
ciation”. In: Zeitschrift fur Experimentelle und Angewandte Psychologie 39.4, 621–645.
O’Regan, J.K. (2009). “Sensorimotor approach to (phenomenal) consciousness”. In: Oxford Com-
panion to Consciousness. Ed. by T. Baynes, A. Cleeremans, and P. Wilken. Oxford: University
Press, pp. 588–593.
O’Regan, J.K. and A. Noë (2001). “A sensorimotor account of vision and visual consciousness”.
In: Behavioral and brain sciences 24.5, 939–972.
Petry, S. (1978). “Perceptual changes during metacontrast”. In: Vision research 18.10, 1337–1341.
Proctor, R. and I. Bernstein (1974). “Metacontrast and simple reaction time: a reexamination”.
In: Journal of Experimental Psychology 102.3, pp. 498–502.
Reeves, A. (1982). “Metacontrast U-shaped functions derive from two monotonic processes”. In:
Perception 11, pp. 415–426.
Roufs, J. A. J. (1974). “Dynamic properties of vision—V: Perception lag and reaction time in
relation to flicker and flash thresholds”. In: Vision Research 14.9, 853–869.
Sackur, J. (2011). “Dynamics of visual masking revealed by second-order metacontrast”. In: Journal
of vision 11.4.
— (2013). “Two dimensions of visibility revealed by multidimensional scaling of metacontrast”. In:
Cognition 126, pp. 173–180.
Sanford, A. J. (1974). “Attention bias and the relation of perception lag to simple reaction time.”
In: Journal of Experimental Psychology 102.3, p. 443.

References
Schiller, P. and M. Smith (1966). “Detection in metacontrast.” In: Journal of Experimental Psy-
chology 71.1, p. 32.
Schneider, G.E. (1969). “Two visual systems.” In: Science 163.3870, pp. 895–902.
Sengpiel, F., A. Sen, and C. Blakemore (1997). “Characteristics of surround inhibition in cat area
17”. In: Experimental Brain Research 116.2, 216–228.
Sherrington, C.S. (1897). “On the reciprocal action in the retina as studied by means of some
rotating discs”. In: Journal of Physiology 21, pp. 33–54.
Sinnett, S. et al. (2007). “A dissociation between visual and auditory hemi-inattention: Evidence
from temporal order judgements”. In: Neuropsychologia 45.3, 552–560.
Spence, C., D.I. Shore, and R. Klein (2001). “Multisensory prior entry.” In: Journal of Experimental
Psychology: General 130.4, p. 799.
Spence, C. and S. Squire (2003). “Multisensory integration: maintaining the perception of syn-
chrony”. In: Current Biology 13.13, R519–R521.
Sperling, G. (1964). “What visual masking can tell us about temporal factors in perception”. In:
Proceeding of the seventeenth international congress of psychology. Amsterdam:Holland.
Steglich, C. and O. Neumann (2000). “Temporal, but not spatial, context modulates a masked
prime’s effect on temporal order judgment, but not on response latency”. In: Psychological
Research 63.1, 36–47.
Sternberg, S. and R.L. Knoll (1973). “The perception of temporal order: Fundamental issues and a
general model”. In: Attention and performance. Kornblum S. (ed). Vol. 4. New York: Academic
Press, pp. 629–685.
Stigler, R. (1910). “Chronophotische Studien über den Umgebungskontrast”. In: Pflügers Archiv
European Journal of Physiology 134.6, 365–435.
Stroud, J.M. (1967). “The Fine Structure of Psychological Time”. en. In: Annals of the New York
Academy of Sciences 138.2, 623–631. issn: 1749-6632.
Tappe, T., M. Niepel, and O. Neumann (1994). “A dissociation between reaction time to sinusoidal
gratings and temporal-order judgment”. In: Perception 23, 335–335.
Van Doorn, H., J. van der Kamp, and G. Savelsbergh (2007). “Grasping the Müller-Lyer illusion:
the contributions of vision for perception in action”. In: Neuropsychologia 45.8, 1939–1947.
Van Gaal, S. and V. A. F. Lamme (2012). “Unconscious High-Level Information Processing Impli-
cation for Neurobiological Theories of Consciousness”. In: The Neuroscientist 18.3, 287–301.
Vroomen, J. et al. (2004). “Recalibration of temporal order perception by exposure to audio-visual
asynchrony”. In: Cognitive brain research 22.1, 32–35.
Waszak, F., P. Cardoso-Leite, and A. Gorea (2007). “Perceptual criterion and motor threshold: a
signal detection analysis of the relationship between perception and action”. In: Experimental
Brain Research 182.2, 179–188.
Waszak, F. and A. Gorea (2004). “A new look at the relationship between perceptual and motor
responses”. In: Visual Cognition 11.8, 947–963.
Yeshurun, Y. and L. Levy (2003). “Transient spatial attention degrades temporal resolution”. In:
Psychological Science 14.3, 225–231.
Zampini, M., D.I. Shore, and C. Spence (2003). “Audiovisual temporal order judgments”. In: Ex-
perimental brain research 152.2, 198–210.
Zampini, M. et al. (2005). “Audio-visual simultaneity judgments”. In: Perception & psychophysics
67.3, 531–544.
D.R.L.Zarebski 41

Additional graphs
Additional graphs
−50 0 50
0.00.20.40.60.81.0
AB
delta
0.00369986
−50 0 50
0.00.20.40.60.81.0
AD
delta
0.00231376
−50 0 50
0.00.20.40.60.81.0
AL
delta
0.000956597
−50 0 50
0.00.20.40.60.81.0
DZ
delta
0.00189996
−50 0 50
0.00.20.40.60.81.0
OA
delta
0.00162351
−50 0 50
0.00.20.40.60.81.0
SL
delta
0.00511933
−50 0 50
0.00.20.40.60.81.0
TC
delta
0.00269202
ProportionofTargetfirst
Figure 22: TOJ: Linear models of selected subjects

Additional graphs
0.00.20.40.60.81.0
AB
−90 −30 30 90
soa
30
60
150
90
120
0.00.20.40.60.81.0
DZ
−90 −30 30 90
soa
30
60
90
120
150
0.00.20.40.60.81.0
TC
−90 −30 30 90
soa
60
90
150
30
120
0.00.20.40.60.81.0
OA
−90 −30 30 90
soa
60
120
150
90
30
0.00.20.40.60.81.0
AD
−90 −30 30 90
soa
30
60
150
120
90
0.00.20.40.60.81.0
SL
−90 −30 30 90
soa
60
90
120
150
30
0.00.20.40.60.81.0
AL
−90 −30 30 90
soa
60
90
120
150
30
proptargetfirst
(a) without motor reponses
0.00.20.40.60.81.0
AB
−90 −30 30 90
soa
30
60
120
150
90
0.00.20.40.60.81.0
DZ
−90 −30 30 90
soa
60
30
90
120
150
0.00.20.40.60.81.0
TC
−90 −30 30 90
soa
60
120
150
30
90
0.00.20.40.60.81.0
OA
−90 −30 30 90
soa
60
150
120
90
30
0.00.20.40.60.81.0
AD
−90 −30 30 90
soa
60
120
30
150
90
0.00.20.40.60.81.0
SL
−90 −30 30 90
soa
60
90
120
150
30
0.00.20.40.60.81.0
AL
−90 −30 30 90
soa
90
120
150
60
30
proptargetfirst
(b) with motor reponses
0.00.20.40.60.81.0
AB
−90 −30 30 90
soaC
no
short
long
0.00.20.40.60.81.0
DZ
−90 −30 30 90
soaC
short
no
long
0.00.20.40.60.81.0
TC
−90 −30 30 90
soaC
short
long
no
0.00.20.40.60.81.0
OA
−90 −30 30 90
soaC
long
short
no
0.00.20.40.60.81.0
AD
−90 −30 30 90
soaC
short
long
no
0.00.20.40.60.81.0
SL
−90 −30 30 90
soaC
long
short
no
0.00.20.40.60.81.0
AL
−90 −30 30 90
soaC
long
short
no
proptargetfirst
(c) without motor reponses
0.00.20.40.60.81.0
AB
−90 −30 30 90
soaC
short
no
long
0.00.20.40.60.81.0
DZ
−90 −30 30 90
soaC
short
no
long
0.00.20.40.60.81.0
TC
−90 −30 30 90
soaC
long
short
no
0.00.20.40.60.81.0
OA
−90 −30 30 90
soaC
no
long
short
0.00.20.40.60.81.0
AD
−90 −30 30 90
soaC
short
long
no
0.00.20.40.60.81.0
SL
−90 −30 30 90
soaC
long
short
no
0.00.20.40.60.81.0
AL
−90 −30 30 90
soaC
long
short
no
proptargetfirst
(d) with motor reponses
Figure 23: TOJ per subjects: x-axis indicate the interval between targets and sounds onsets (∆)
and y-axis represent the proportion of ”Target ﬁrst” answers. Figs.23(a) and 23(b) detail the
diﬀerent SOA in comparison with the unmasked condition. Figs.23(c) and 23(d) group conditions
in Long SOA –90, 120,150 ms– Short SOA –30,60ms– and Unmasked conditions
D.R.L.Zarebski 43

Additional graphs
600 800 1000 1200 1400
60080010001400
no
150
0
120
90
60
30
AB
600 800 1000 1200 1400
60080010001400
no
60
30
120
90
150
0
DZ
600 800 1000 1200 1400
60080010001400
no
0
30
60
120
90
150
TC
600 800 1000 1200 1400
60080010001400
no
0
120
30
150
60
90
OA
600 800 1000 1200 1400
60080010001400
no
0
120
150
60
30
90
AC
600 800 1000 1200 1400
60080010001400
no
60
0
120
30
150
90
AD
600 800 1000 1200 1400
60080010001400
no
0
120
60
90
30
150
SL
600 800 1000 1200 1400
60080010001400
no
90
0
150
30
60
120
AL
Delay (ms)
EstimDelays(ms)
(a) without motor reponses
600 800 1000 1200 1400
60080010001400
no
120
30
150
60
0
90
AB
600 800 1000 1200 1400
60080010001400
no
0
150
60
30
120
90
DZ
600 800 1000 1200 1400
60080010001400
no
0
150
120
30
90
60
TC
600 800 1000 1200 1400
60080010001400
no
90
30
60
120
0
150
OA
600 800 1000 1200 1400
60080010001400
no
120
30
90
0
150
60
AC
600 800 1000 1200 1400
60080010001400
no
150
60
120
0
90
30
AD
600 800 1000 1200 1400
60080010001400
no
60
90
30
0
150
120
SL
600 800 1000 1200 1400
60080010001400
no
150
60
120
90
30
0
AL
Delay (ms)
EstimDelays(ms)
(b) with motor reponses
600 700 800 900 1100
60080010001200
long
short
no
AB
600 700 800 900 1100
60080010001200
short
no
long
DZ
600 700 800 900 1100
60080010001200
short
no
long
TC
600 700 800 900 1100
60080010001200
no
long
short
OA
600 700 800 900 1100
60080010001200
no
short
long
AC
600 700 800 900 1100
60080010001200
short
no
long
AD
600 700 800 900 1100
60080010001200
no
long
short
SL
600 700 800 900 1100
60080010001200
long
no
short
AL
Delay (ms)
EstimDelays(ms)
(c) without motor reponses
600 700 800 900 1100
60080010001200
long
no
short
AB
600 700 800 900 1100
60080010001200
short
long
no
DZ
600 700 800 900 1100
60080010001200
short
no
long
TC
600 700 800 900 1100
60080010001200
no
short
long
OA
600 700 800 900 1100
60080010001200
long
short
no
AC
600 700 800 900 1100
60080010001200
no
short
long
AD
600 700 800 900 1100
60080010001200
no
short
long
SL
600 700 800 900 1100
60080010001200
long
no
short
AL
Delay (ms)
EstimDelays(ms)
(d) with motor reponses
Figure 24: estim per subjects: ﬁgs.24(a) and 24(b) detail the diﬀerent SOA in comparison with the
unmasked condition. Figs.24(c) and 24(d) group conditions in Long SOA –90, 120,150 ms– Short
SOA –30,60ms– and Unmasked conditions

Additional graphs
0 200 400 600 800
0.0000.0020.0040.0060.0080.010
AB
N = 1995 Bandwidth = 7.788
0 200 400 600 800
0.0000.0020.0040.0060.0080.010
AC
N = 937 Bandwidth = 14.19
0 200 400 600 800
0.0000.0020.0040.0060.0080.010
AD
N = 1913 Bandwidth = 7.113
0 200 400 600 800
0.0000.0020.0040.0060.0080.010
AL
N = 1738 Bandwidth = 19.18
0 200 400 600 800
0.0000.0020.0040.0060.0080.010
DZ
N = 1917 Bandwidth = 13.48
0 200 400 600 800
0.0000.0020.0040.0060.0080.010
OA
N = 1891 Bandwidth = 12.33
0 200 400 600 800
0.0000.0020.0040.0060.0080.010
SL
N = 1907 Bandwidth = 7.562
0 200 400 600 800
0.0000.0020.0040.0060.0080.010
TC
N = 1941 Bandwidth = 10.34
Density
Figure 25: Distributions of RT per subjects
D.R.L.Zarebski 45

memoire_zarebski_2013

More Related Content

What's hot (19)

Viewers also liked (8)

Similar to memoire_zarebski_2013 (20)

memoire_zarebski_2013