Exp Brain Res (2009) 197:153–161

DOI 10.1007/s00221-009-1901-7

RESEARCH ARTICLE

Can illusory deviant stimuli be used as attentional distractors to record vMMN in a passive three stimulus oddball paradigm?

Maria Flynn · Alki Liasis · Mark Gardner · Stewart Boyd · Tony Towell

Received: 23 September 2008 / Accepted: 9 June 2009 / Published online: 24 June 2009© Springer-Verlag 2009

Abstract A passive three stimulus oddball paradigm wasused to investigate Visual Mismatch Negativity (vMMN) acomponent of the Event Related Potential (ERP) believedto represent a central pre-attentive change mechanism.Responses to a change in orientation were recorded tomonochrome stimuli presented to subjects on a computerscreen. One of the infrequent stimuli formed an illusoryWgure (Kanizsa Square) aimed to capture spatial attentionin the absence of an active task. Nineteen electrodes (10–20system) were used to record the electroencephalogram infourteen subjects (ten females) mean age 34.5 years. ERPsto all stimuli consisted of a positive negative positive com-plex recorded maximally over lateral occipital areas. Thenegative component was greater for deviant and illusorydeviant compared to standard stimuli in a time window of170–190 ms. A P3a component over frontal/central elec-trodes to the illusory deviant but not to the deviant stimulussuggests the illusory Wgure was able to capture attentionand orientate subjects to the recording. Subtraction wave-forms revealed visual discrimination responses at occipitalelectrodes, which may represent vMMN. In a control studywith 13 subjects (11 females; mean age 29.23 years), using

an embedded active attention task, we conWrmed the exis-tence of an earlier (150–170 ms) and attenuated vMMN.Recordings from an intracranial case study conWrmed sepa-ration of N1 and discrimination components to posteriorand anterior occipital areas, respectively. We conclude thatalthough the illusory Wgure captured spatial attention in itsown right it did not draw suYcient attentional resourcesfrom the standard–deviant comparison as revealed whenusing a concurrent active task.

Keywords Event-related potential · Visual mismatch negativity · Kanizsa Wgure · Orientation

Introduction

The Mismatch Negativity (MMN) is deWned as a compo-nent of the Event Related Potential (ERP) that can beevoked to stimulus change in the absence of attention. TheMMN is usually elicited when a deviant stimulus is pre-sented within a sequence of standard stimuli. The auditoryMMN has been identiWed as a negative deXection usuallypeaking at 150–200 ms from change onset and is related toautomatic discrimination processing and sensory memorymechanisms (see Näätänen et al. 2005, 1997; Schröger1997, for reviews).

A number of studies have identiWed visual MismatchNegativity (vMMN), as a negative deXection 100–250 mspost-stimulus change onset (see Pazo-Alvarez et al. 2003for a review). These studies have reported vMMN either to‘match’ and ‘non-match’ tasks where the stimuli are pre-sented with equiprobability to control the eVects of globalstimulus presentation (Fu et al. 2003) for changes in orien-tation and spatial frequency (Kimura et al. 2006) forchanges in spatial frequency), or within an oddball

M. Flynn · A. Liasis · M. Gardner · T. Towell (&)Department of Psychology, University of Westminster, 309 Regent Street, London W1B 2UW, UKe-mail: towella@wmin.ac.uk

A. LiasisDepartment of Ophthalmology, Great Ormond Street Hospital for Children, London WC1N 3JH, UK

S. BoydDepartment of Clinical Neurophysiology, Great Ormond Street Hospital for Children, London WC1N 3JH, UK

123

154 Exp Brain Res (2009) 197:153–161

paradigm, whereby a variety of dimensions of the visualstimulus that are known to be important in early visualprocessing are manipulated. These include changes in spa-tial frequency (Maekawa et al. 2005), motion (Kremlaceket al. 2006), colour (Czigler et al. 2004; Czigler et al. 2002;Horimoto et al. 2002), form (Berti and Schroger 2004;Besle et al. 2005; Stagg et al. 2004) and orientation(Astikainen et al. 2004, 2008; Czigler and Csibra 1992).

Similar to the auditory MMN the vMMN is thought toreXect the memory based detection of deviant stimuli ratherthan refractoriness (see Czigler et al. 2007 for a detaileddiscussion). However, in contrast to the correlation seenbetween auditory MMN and behavioural detection of devi-ants (Winkler et al. 1993) there appears to be no such rela-tionship in the visual modality. The amplitude of vMMNdoes not increase beyond 40 ms stimulus onset asynchrony(SOA) of a masking stimulus whilst detection performanceof deviant stimuli and RT improve up to 174 ms SOA(Czigler et al. 2007). These Wndings strongly suggest thatovert detection of visual deviance is not the sole mechanismunderlying vMMN.

In comparison to other ERP change components, such asN2 and P3, the MMN can be elicited in the absence ofattention (Pazo-Alvarez et al. 2003). Therefore, in order todiVerentiate between MMN and other ERP change compo-nents the subjects’ attention is typically drawn away fromthe test stimuli, employing a variety of behavioural tasks.For example, Stagg et al. (2004) and Tales et al. (1999)required participants to press a button in response to targetstimuli, Astikainen et al. (2004) used an auditory distrac-tion task whereby participants were required to focus theirattention on counting the number of words in a story whilstbeing presented with visual stimuli.

It has generally been understood that a concurrent activetask is mandatory in eliciting vMMN to control for theeVects of attention so that resources are allocated awayfrom the standard–deviant discrimination towards theactive task (Heslenfeld 2003; Czigler 2007). However, notall patient populations can meet the demands of an activetask. Therefore, in the present study a three-stimulus pas-sive oddball paradigm was developed. Stimuli diVered withregard to orientation of local endline type pacman Wguresand their information/entropy content. So in addition tostandard and deviant stimuli, an infrequent illusory deviantstimulus was introduced in order to investigate the eVectsof attention. The illusory deviant stimulus was a KanizsaWgure (Kanizsa 1976) which formed an illusory square, asalient event thought to demand attention to reconstructcontours that are absent from visual images (Kaiser et al.2004).

A consistent Wnding in ERP research is that the P3 wave,a positive deXection occurring from 280 to 400 ms post-stimulus indicates attentional processing (see Hruby and

Marsalek 2003; Polich 2003; Hagen et al. 2006 forreviews). The P3 can be further divided into the subcompo-nents P3a and P3b. P3a originates from frontal attentionmechanisms to task novelty and/or distractors whilst theP3b is generated in more temporal/parietal regions and isassociated with context updating and memory storage oper-ations (Polich 2007). We therefore set out to validate theuse of illusory deviant stimuli in orienting attention in apassive and active task in the context of a vMMN para-digm. It was predicted that in the passive paradigm with notask conditions that discrimination components possiblyreXecting MMN would be evoked by both the deviant andillusory deviant stimuli while a P3a component would onlybe evident to illusory deviant stimuli that captured attention.

To test the use of the illusory deviant stimulus in orien-tating attention from the standard–deviant discrimination acontrol study was carried out containing an embeddedactive task. Last, the generator sources of the visual ERPcomponents were explored using intracranial recordings ina subject undergoing presurgical evaluation for epilepsysurgery.

Methods

Participants

Study 1 and 2

With ethical approval and informed consent 14 healthyadults (mean age 34.5 § 8.6 years (10 females) wererecruited for study 1 and 13 healthy adults (mean age29.23 § 8.8 years (11 females) for study 2. Subjectsreported no history of neurological disease and had normalor corrected-to-normal visual acuity.

Study 3

With hospital ethical approval and patient and parental con-sent, a 15-year-old male with focal epilepsy undergoingpre-surgical evaluation for resection of a R anterior parietallesion provided the opportunity to examine whether therewas a dissociation of detection and discrimination compo-nents of the visual ERP.

Stimuli and procedure

Study 1

Three monochrome endline type stimuli based on pacmanWgures were employed in a behaviourally silent oddballparadigm where the ratio of standards to deviants and illu-sory deviants was 8:1:1. The stimuli in (Fig. 1a), diVered

123

Exp Brain Res (2009) 197:153–161 155

from each other only in terms of the orientation of elements(which were oriented unsystematically around their axesfor the standard and deviant stimuli and formed an illusoryKanizsa Wgure for the illusory deviant stimulus. The stimuliwere generated employing STIM software (Neuroscan-STIM version 4; Compumedics USA, Ltd., El Paso, TX,USA) and presented on a computer screen subtending 4°.The stimuli appeared on the screen for 400 ms with aninter-stimulus interval of 600 ms. Subjects were seatedcomfortably in a darkened room 1 m away from the screenand requested to Wxate on a small red dot in the centre ofthe screen that was present throughout recording. Withinthe oddball paradigm stimuli were presented in a pseudo-random sequence ensuring that deviant and illusory deviantstimuli were interspersed with standard stimuli. In study 1,the stimuli were presented in Wve blocks of 225 stimuli withup to a minute break between blocks. At the end of the odd-ball recording blocks of 64 deviants and illusory deviants‘alone’ were presented.

Study 2

The same stimuli and procedure as in Study 1 were utilizedwith the exception that an active attention task was embed-ded in the three stimulus oddball paradigm. Within theblocks of 225 stimuli, during the interstimulus interval(ISI), a small red square replaced the small red Wxation doton 22 trials chosen at random. The red square appeared atthe start of the 600 ms ISI and stayed on the screen for200 ms. Subjects were instructed to focus their attention onthe red Wxation dot and press the right button of a mouse asquickly as possible whenever the red square appeared.Inclusion criteria were based on participants achieving 90%or more correct responses, excluding false positives.

In study 3, the two blocks of the stimuli were presentedwith no embedded active attention task.

Electroencephalogram recording and data analysis

Study 1 and 2

Nineteen silver–silver chloride electrodes were used torecord the electroencephalogram (EEG) activity and werepositioned at sites in accordance with the International 10–20 system (Fz, F3, F4, Cz, C3, C4, T3, T4, Pz, P3, P4, Oz,O1, O2, T5, T6, VEOG, M1, M2). The reference electrodeand the ground electrode were placed at the right and leftmastoid, respectively. An electrode was placed above theleft eye to enable online artefact rejection of eye blinks.Continuous EEG was collected using Neuroscan-SCANversion 4.3; Compumedics USA, Ltd., El Paso, TX, USA ata sampling rate of 1,000 Hz, with a low pass of 100 Hz anda high pass of 1 Hz and stored on a computer for oZineanalysis.

Continuous EEG data were epoched oZine ¡100 mspre-stimulus to +500 ms post-stimulus. The epochs weredigitally Wltered with a band pass 1–30 Hz and baseline cor-rected. Epochs containing transients greater than §150 �Vwere excluded from further analysis. For each subject, ERPswere averaged separately for standard, deviant and illusorydeviant stimuli employing Fz as a reference and grand aver-age waveforms were constructed. Additional ERPs wereconstructed in study 2 for the red Wxation dot and for the redsquare that replaced the Wxation dot on a number of trials.ERPs to standard stimuli were constructed from epochs thatpreceded deviant stimuli. As in previous studies (Stagg et al.2004; Tales et al. 1999), averaged mastoids were employedas a reference to investigate P3 activity.

Fig. 1 a Stimuli presented in oddball paradigm with pseudo-random sequence of 8:1:1, respectively. (i) standard, (ii) deviant and (iii) illusory deviant forming a Kanizsa square. b, c Grand average waveforms referenced to Fz at O1 and O2, respectively for (i) Standard (dashed line), deviant (solid line) and illusory deviant stimuli (dotted line) (ii) Deviant minus standard (iii) Illusory deviant minus standard. Note the dis-crimination responses in ii) and iii) with an additional negative component in (iii) corresponding to an inverted P3

a) b) -8µV c)N1

P1 P2

O1 O2i)

ii)

iii)

100ms

123

156 Exp Brain Res (2009) 197:153–161

From the grand average waveforms MMN-like diVer-ences were identiWed on the basis of known negative polar-ity, known emergence over posterior electrode positionsand typical latency range (100–250 ms post-stimulus: Pazo-Alvarez et al. 2003). In each study, the maximal diVerencebetween ERPs to standards and deviants was identiWed atoccipital sites and a 20 ms time window was centred at thislatency for electrodes P3, P4, O1, O2, T5, T6 (Astikainenet al. 2008). Mean amplitudes for the time windows werecalculated relative to the mean voltage of a 100 ms pre-stimulus baseline for each participant for the standard, devi-ant and illusory deviant stimuli. The mean amplitudes wereanalysed using ANOVA. In addition, subtraction wave-forms were constructed of deviant minus standard and illu-sory deviant minus standard.

Study 3

The patient was implanted with a 32-contact sub-dural plat-inum grid straddling the parietal and pre-motor gyri and a6-contact strip extending posteriorly over the inferior parie-tal cortex such that the most distal contact (S1) overlay theR occipital cortex (Fig. 3a).

Results

Study 1

A visual response was recorded in all subjects in all trialsconsisting of a P1–N1–P2 waveform. Grand average wave-forms were constructed for the standard, deviant and illu-sory deviant stimuli (see Fig. 1 for waveforms at O1 andO2). The maximal diVerence between ERPs to standardsand deviants was at approximately 180 ms post-stimulus atoccipital electrodes. A 20 ms time window was centred atthis latency for electrodes O1, O2, P3, P4, T5, T6 and, foreach participant, mean amplitudes for this time window cal-culated relative to the mean voltage of a 100 ms pre-stimu-lus baseline for standards, deviants and illusory deviants.Mean amplitudes and standard deviations for the standard,deviant and illusory deviant are shown in Table 1.

A three-way within subjects ANOVA was used to ana-lyse the mean amplitude data in the 170–190 ms time win-dow. Pairwise comparison of means was carried out usingbonferroni corrected t tests. Factors were location (occipi-tal, parietal, temporal), hemisphere (left, right) and stimulus(standard, deviant and illusory deviant). The amplitudediVered signiWcantly with location [F(2,26) = 11.880;P < 0.001] and stimulus type [F(2,26) = 15.886; P < 0.001]but not with hemisphere [F(1,13) = 0.233; P = 0.794].There was a statistically signiWcant interaction betweenlocation and stimulus [F(4.52) = 6.503; P = 0.001,

P < 0.001], indicating that the amplitude of the deviantstimulus was greater than the standard stimulus at occipital(t = 4.004; df = 13; P = 0.002) and temporal electrodes(t = 4.552; df = 13; P = 0.001) and that the amplitude of theillusory deviant was greater than the standard at occipital(t = 4.507; df = 13; P = 0.001), temporal (t = 4.552;df = 13; P = 0.001) and parietal electrodes (t = 4.276;df = 13; P = 0.001).

DiVerence waveforms of deviant minus standard andillusory deviant minus standard both revealed vMMN com-ponents (Fig. 1b, c). When comparing the deviant to thestandard ERP, using the point-by-point t test algorithm(P < 0.05; one-tailed) against baseline there were signiW-cant diVerences at O1 between 173 and 217 ms (181–203 ms; P < 0.01) and at O2 between 178 and 208 ms(185–196 ms); P < 0.01). Comparing the illusory deviant tothe standard ERP against baseline, there were signiWcantdiVerences (P < 0.05; one-tailed) at O1 between 164 and212 ms (175–203 ms; P < 0.01) and at O2 between 169 and212 ms (181–199 ms; P < 0.01).

Illusory deviant stimuli evoked an additional late nega-tive component at 234 ms at Oz. To examine whether thiscomponent corresponded to an inverted P3 component thewaveforms were re-referenced to averaged mastoids. Wewere able to reveal a positive component over the fronto-central electrode sites corresponding to P3a. At Fz thiscomponent had an onset latency of 244 ms, SD = 13 msand a peak latency of 290 ms, SD = 27 ms with a peakamplitude of 4.19 �V, SD = 2.06 �V.

To examine whether the diVerences observed in the sub-traction waveforms were confounded by pure stimulusdiVerences we compared the discrimination waveform tothe deviant stimulus to the discrimination waveform whenthat same stimulus was presented alone, i.e. out of contextand not in an oddball paradigm. Point-by-point t testsrevealed no signiWcant diVerences between the deviant–standard and deviant alone-deviant waveforms suggesting

Table 1 Mean ERP amplitude (�V) and standard deviation (SD) foreach stimulus type at electrode sites for the 170–190 ms time windowfor the passive task (n = 14)

Electrode site

Mean amplitude (�V) and standard deviation (§SD)

Stimulus

Standard Deviant Illusory deviant

O1 2.98 § 2.32 5.67 § 3.36 7.10 § 4.10

O2 3.11 § 2.73 5.66 § 3.38 7.33 § 4.38

P3 1.99 § 1.68 3.75 § 2.85 5.03 § 3.49

P4 1.90 § 1.91 3.43 § 2.27 5.08 § 3.12

T5 2.36 § 2.03 4.59 § 2.75 5.42 § 3.07

T6 2.62 § 1.83 4.70 § 2.24 5.13 § 2.75

123

Exp Brain Res (2009) 197:153–161 157

that when the deviant stimulus was presented alone and outof context it behaved in a similar way to the standard stimu-lus even though it was physically diVerent. The same proce-dure was used to compare the illusory deviant stimulus inthe context of an oddball paradigm with the illusory deviantstimulus presented alone. Similarly, there were no signiW-cant diVerences between the illusory deviant–standard andillusory deviant alone-illusory deviant waveforms.

Study 2

As in study 1, a visual response was recorded for all sub-jects consisting of a P1–N1–P2 waveform. Grand averagewaveforms were constructed for the standard, deviant andillusory deviant stimuli (see Fig. 2 for waveforms at O1 andO2). The maximal diVerence between ERPs to standardsand deviants was at approximately 160 ms post-stimulus atoccipital sites. A 20-ms time window was centred at thislatency for electrodes O1, O2, P3, P4, T5, T6 and meanamplitudes for this time window calculated relative to the

mean voltage of a 100 ms pre-stimulus baseline for stan-dards, deviants and illusory deviants for each participant.Mean amplitudes and standard deviations for the standard,deviant and illusory deviant are shown in Table 2.

A three-way within subjects ANOVA was used to ana-lyse the mean amplitude data of the 150–170 ms time win-dow. Pairwise comparison of means was carried out usingbonferroni corrected t tests. Factors were location (occipi-tal, parietal, temporal), hemisphere (left, right) and stimulus(standard, deviant and illusory deviant). The amplitudediVered signiWcantly with location [F(2,24) = 16.874;P < 0.001], hemisphere [F(2,24) = 7.059; P = 0.021] andstimulus type [F(2,24) = 14.254; P < 0.001]. There was asigniWcant interaction between location and stimulus[F(4.48) = 10.636; P < 0.001] indicating that the amplitudeof the deviant stimulus was greater than the standard stimu-lus at occipital (t = 3.796; df = 12; P = 0.003) and temporal(t = 3.147; df = 12; P = 0.008) electrodes. The amplitude ofthe illusory deviant stimulus was greater than the standardstimulus at occipital (t = 4.494; df = 12; P = 0.001), tempo-ral (t = 4.425; df = 12; P = 0.001) and parietal (t = 4.105;df = 12; P = 0.001) electrodes. There was a signiWcantinteraction between hemisphere and stimulus [F(2,24) =3.402; P = 0.050] indicating that in the left hemisphere themean amplitude was greater for the deviant (t = 4.194;df = 12; P = 0.001) and illusory deviant (t = 5.536; df = 12;P < 0.001) than for the standard. In the right hemisphere themean amplitude of the illusory deviant (t = 5.944; df = 12;P < 0.001) was greater than the standard as was the deviantbut to a lesser extent (t = 2.952; df = 12; P = 0.012).

DiVerence waveforms of deviant minus standard andillusory deviant minus standard both revealed attenuatedvMMN components (Fig. 2b, c). When comparing the devi-ant to the standard ERP, using the point-by-point t testalgorithm (P < 0.05; one-tailed) against baseline, therewere signiWcant diVerences at O1 between 138 and 176 msbut no signiWcant diVerences were apparent at O2. When

Fig. 2 a Grand average waveforms referenced to Fz for standard(dashed line), deviant (solid line) and illusory deviant stimuli (dottedline) at O1 and O2. Note the P3a component seen only to illusory devi-ant stimuli. b Deviant minus standard. c Illusory deviant minus stan-dard. d Grand average waveforms for the rarely occurring red Wxationsquare (solid line) and for the central Wxation dot (dotted line). Note:the attenuated vMMN in b and the P3b wave to the task in d

b)

a)

O2O1

d)

c)

-8µV

100ms

N1

P1P2

P3

Table 2 Mean ERP amplitude (�V) and standard deviation for eachstimulus type at electrode sites at 150–170 ms for the active task(n = 13)

Electrode site

Mean amplitude (�V) and standard deviation (§SD)

Stimulus

Standard Deviant Illusory deviant

O1 4.12 § 2.22 5.12 § 2.10 7.24 § 2.79

O2 5.20 § 3.40 6.25 § 3.43 9.27 § 5.20

P3 2.32 § 1.76 2.79 § 1.83 3.51 § 2.22

P4 3.50 § 2.33 4.07 § 2.01 5.56 § 3.28

T5 3.23 § 1.75 4.18 § 1.82 5.20 § 2.00

T6 4.90 § 2.35 5.6 4 § 2.36 7.48 § 3.92

123

158 Exp Brain Res (2009) 197:153–161

comparing the illusory deviant to the standard ERP againstbaseline, there were signiWcant diVerences (P < 0.05; one-tailed) at O1 between 142 and 178 ms and at O2 between147 and 177 ms.

Illusory deviant stimuli evoked an additional late nega-tive component at occipital electrodes. When re-referencedto averaged mastoids this component was positive over thefronto-central electrode sites. At Fz this had an onsetlatency of 223 ms, SD = 18 ms and a peak latency of282 ms, SD = 22 ms with a peak amplitude of 5.17 �V,SD = 2.72 �V.

All participants completed the active task (pressing themouse when the red Wxation dot was replaced with a redWxation square) within the limits of the inclusion criteria.As expected, the active task evoked a P3b componentshowing that the participants’ attention was engaged withthe task (See Fig. 2d).

Study 3: intracranial recording

A negative positive negative complex was recorded maxi-mally to all stimuli at the most posterior electrode site (S1)(Fig. 3b). The latency and amplitude of the Wrst major neg-ative component (N1) was similar for standard, deviantstimuli and illusory deviant stimuli (153 ms and ¡32.39 �V,153 ms and ¡48.54 �V, 162 ms and ¡45.50 �V, respec-tively). Responses to stimulus discrimination (visual

mismatch) were recorded more anteriorly at S3 and S4 andwere characterised by enhanced positivities at about 90 ms,42.32 �V and 219 ms, 87.21 �V for S3 and enhancedpositivities at 88 ms, 28.55 �V and 237 ms, 45.93 for S4,either side of the major negative component (Fig. 3b).Subtraction waveforms (deviant–standard) revealed dis-crimination responses with positive peaks at 86 and 219 msfor electrode S3 and 93 and 233 ms at electrode S4. Inaddition, a later positive response to the illusory deviantstimulus was seen at around 386 ms over pre-motor regions(G9 > G17 and G18), suggesting activation of a frontal sys-tem to stimulus novelty and/or target detection (Fig. 3c).

Discussion

The main result from this study is that visual discriminationresponses including vMMN components have beenrecorded in a behaviourally silent oddball paradigm to achange in orientation. The stimuli utilised in this studyevoked a response that was more negative to the deviantstimuli than to the standard stimuli in the period 150–200 ms after stimulus onset. Whilst we acknowledge thatthere were physical diVerences between the stimuli, andthat these changes were not equal between standard–devi-ant and standard–illusory deviant comparisons, the employ-ment of a ‘deviant alone’ and ‘illusory deviant alone’conditions served as controls. Subsequent subtractionwaveforms using the subtraction method suggested byKraus et al. (1995) for delineating the MMN reveals thatthe diVerence in negativity was attributable not to physicaldiVerences in the stimuli themselves, but by the context inwhich the stimuli were presented.

The presence of a P3a over frontal/central electrodes forthe illusory deviant grand average waveform but not for thestandard or deviant grand average waveforms, suggests thatthe Kanizsa square captured attention. Without the controltask this would imply that the enhanced negativity exhib-ited by the deviant compared to the standard may notdepend on attention. However, the use of the illusory Wgureis supported by Senkowski et al. (2005) who found thatKanizsa Wgures automatically capture spatial attentionwhen used as visual cues and Wallach and Slaughter (1988)who found that the familiarity of the illusory shapeincreases the likelihood that the shape will be perceived. Inaddition, a number of clinical studies show that theresponse to Kanizsa Wgures is robust. In an ERP study,Grice et al. (2003) examined perceptual completion in par-ticipants with Williams Syndrome—a genetic disorder inwhich visuo-spatial performance is poor, and found thatalthough the underlying neural mechanisms of the partici-pants with Williams Syndrome may be diVerent to controls,their ability to perceive illusory contours was apparently

Fig. 3 a Co-registered sub-dural electrode locations, dotted eclipsedenotes surface visible lesion, x seizure onset zone, + somatosensoryERP localised. b Standard and deviant waveforms from the six stripcontacts—dashed waveform represents the standard ERP, solid wave-form the deviant. At S1 and S2 the illusory deviant waveform did notdiVer from the deviant or standard and no consistent changes were seenat S3 and S4. For reasons of clarity the illusory deviant waveform is notshown. Peak amplitude of the N1 and inverted discrimination compo-nent shown by the shaded area of the waveform. c Anterior grid elec-trodes revealing a later positive response to the illusory deviantstimulus shown by the shaded area of the waveform. Dashed, solid anddotted lines represent ERPs to standard, deviant and illusory deviantstimuli, respectively

G9

G17

G18

s1

s2

s3

s4

s5

s6

100ms

-30uV

b)

S1S6

G17G9

G32

G1

x+

x+a)

c)

123

Exp Brain Res (2009) 197:153–161 159

normal. Milne and Scope, in their 2007 study, suggestedthat the perception of illusory contours in participants withAutistic Spectrum Disorder was intact.

Observation and statistical analysis of the waveformsand the discrimination components reveals that the ampli-tude component N1 for the illusory deviant at the lateraloccipital electrodes was greater than for the standard ordeviant stimuli. Many previous studies have demonstratedan enhanced visual N1 amplitude component to attended-location stimuli (see Vogel and Luck 2000 for a review)and this evidence further suggests that the illusory deviantstimulus captured attention whilst the standard and deviantstimuli did not.

As with several previous studies (e.g. Tales et al. 1999)we also engaged an active control task that required partici-pants to press a button at the occurrence of a change inshape of the central Wxation dot. The understanding here isthat attentional resources are drawn from the standard–deviant discrimination to the active task. Under these con-ditions we were able to conWrm the existence of vMMNresponses although they were signiWcantly reduced inamplitude. The underlying mechanisms of vMMN are stillto be resolved although a number of studies have suggesteda memory based rather than refractoriness explanation (seeCzigler 2007). Such a visual based memory system wouldrely on the representation of regularity following repeatedexposure to identical frequent stimuli. The violation of suchregularity following the presentation of a deviant stimuluswould elicit an enhanced posterior negativity commonlyseen as vMMN in subtraction waveforms. However, in thismodel, it appears that longer sequences (10–15) of fre-quent/standard and identical unattended stimuli will inXu-ence the generation of vMMN (Czigler and Pato 2009). Inthe current study, the median number of continuous stan-dard sequences was 4 and this may account for the rela-tively low amplitude of the vMMN responses in study 2.The latency of the responses in the current study are consis-tent within the general window for vMMN responses ofbetween 100 and 250 ms (Pazo-Alvarez et al. 2003)although it is known that latency and duration of vMMNwill diVer according to stimulus characteristics and taskcomplexity with less salient changes and more complexrules resulting in longer latency and less phasic vMMNresponses (Czigler et al. 2006).

Previous studies on vMMN have tended to engage activetasks embedded in more peripheral areas of the visual Weldand one study speciWcally set out to assess the contributionof the magnocellular system (Kremlacek et al. 2006). Thispathway forms the dorsal stream and is not sensitive to col-our or detail but is thought to be responsible for pre-atten-tive detection of motion stimuli. Whilst in the present studywe cannot exclude the contribution of the magnocellularsystem, our Wndings of a vMMN in the macular Weld at 4°

of arc also reveals the contribution of the parvocellular sys-tem and ventral stream in detecting diVerences in thesequence of unattended central stimuli. The parvocellularsystem is particularly adapted to colour and high-contrastblack and white detailed information.

Besle et al. (2005) using the deformation of a circle as adeviant stimulus embedded within an active task presentedwithin 2° of arc demonstrated bilateral vMMN responses at216 ms being maximal at electrode PO3 and PO4. In anactive geometric shape discrimination task P1, N1 and P2components were identiWed at 80, 140 and 200 ms, respec-tively, and the N1 and P2 components became less sharpand more diVuse as stimulus presentation changed between4°, 8° and 12° of arc (Shoji and Ozaki 2006).

Extra deviant stimuli conceptulised as distractor stimulihave also been used in the auditory modality to manipulateattention. For instance, Schroger et al. (2000) and Schrogerand WolV (1998) in an auditory duration discriminationtask found that task irrelevant distractors in the form ofsmall changes in frequency prolonged reaction times andelicited MMN and P3a components, reXecting orientationtowards the distractor.

Recordings from the intracranial case study support theseparation of detection and discrimination processes withinthe visual cortex. The N1 component at 153 ms located atthe more posterior electrodes corresponds to the scalprecorded N1 at 167 ms. The waveforms were similar for thestandard, deviant and illusory deviant stimuli. However, atadjacent posterior electrodes (S3 and S4) the deviant stim-uli evoked early and later positivities that probably contrib-ute to the scalp recorded MMN. With respect to scalprecordings these potentials to stimulus discrimination areinverted in polarity and the Wrst positive component is seenrelatively early at around 90 ms. These Wnding are consis-tent with an MEG study showing strong activation of thelateral occipital cortex at around 155 ms post-stimulus(Halgren et al. 2003). In MEG studies comparison of illu-sory Kanizsa stimuli with control stimuli reveals activationbetween 100 and 350 ms post-stimulus (Kaiser et al. 2004)and at around 280 ms (Halgren et al. 2003). It is believedthat illusory contour sensitivity may Wrst occur in middle tohigher order visual processing areas and that feedbackmodulation from lateral occipital areas will activate V1 andV2 areas (Kaiser et al. 2004).

Early cortical processing in the visual cortex has alsobeen reported from MEG studies using Xash stimuli thatreach medial occipital areas around 47 ms (Inui and Kakigi2006). Recent studies using intracranial recordings demon-strate activation in the superior parietal lobule at 75 ms tocoloured disc stimuli presented in the macular Weld(Molholm et al. 2006) and recordings from the striatecortex to alternating stimuli have been reported as a P55followed by a more consistent N75 (Farrell et al. 2007).

123

160 Exp Brain Res (2009) 197:153–161

Polarity inversions between the cortex and the scalp canindicate local generator sources in that region of cortex. Asthese scalp recorded N1 and MMN Welds are interactions ofthe super imposition of several bilateral generators it isdiYcult to understand how focal intracranial potentials con-tribute to the scalp recorded N1 and MMN. The invertedbiWd positive discrimination component may well representthe existence of one or more local generator sources tochange detection. Complex and widespread activation hasalso been recorded to alternating and on/oV stimuli from thestriate cortex and visual association areas (Farrell et al.2007) further supporting the view that it is diYcult toentangle the generator sources that contribute to responsesmeasured at the scalp.

Intracerebral potentials to rare distractor visual and audi-tory stimuli have been recorded from frontal regions as awidespread negative–positive–negative waveform atapproximate latencies of 210–280–390 ms, respectively(Baudena et al. 1995). It is believed that this waveform cor-responds with the scalp recorded N2a/P3a/slow wave that isassociated with orienting. In the present study the later pos-itivity to the illusory deviant stimulus seen at around386 ms over pre-motor regions may correspond to this nov-elty orienting process.

In conclusion, we suggest that visual discriminationpotentials containing vMMN components can be elicitedusing a paradigm with no task demands. The inclusion ofan illusory square was intended to capture the subject’sattention and therefore orientate them to the recording. Theexistence of attenuated vMMN when subjects engaged inan active distractor task supports the contention that theillusory square was unable to command all resources awayfrom the standard–deviant comparison.

Acknowledgments We are grateful to the anonymous referees fortheir critical comments and suggestions for improvements on earlierversions of the manuscript.

References

Astikainen P, Ruusuvirta T, Wikgren J, Korhonen T (2004) The humanbrain processes visual changes that are not cued by attended audi-tory stimulation. Neurosci Lett 368:231–234

Astikainen P, Lillstrang E, Ruusuvirta T (2008) Visual mismatch neg-ativity for changes in orientation—a sensory memory-dependentresponse. Eur J Neurosci 28:2319–2324

Baudena P, Halgren E, Heit G, Clark JM (1995) Intracerebral potentialsto rare target and distractor auditory and visula stimuli. III. Frontalcortex. Electroencephalogr Clin Neurophysiol 94:251–264

Berti S, Schroger E (2004) Distraction eVects in vision: behaviouraland event-related potential indices. Neuroreport 15:665–669

Besle J, Fort A, Giard MH (2005) Is the auditory sensory memory sen-sitive to visual information. Exp Brain Res 166:337–344

Czigler I (2007) Visual mismatch negativity. Violation of nonattendedenvironmental regularities. J Psychophysiol 21:224–230

Czigler I, Csibra G (1992) Event-related potentials and the identiWca-tion of deviant visual stimuli. Psychophysiology 29:471–485

Czigler I, Pato L (2009) Unnoticed regularity violation elicts change-related brain activity. Biol Psychol 80:339–347

Czigler I, Balázs L, Winkler I (2002) Memory-based detection of task-irrelevant visual changes. Psychophysiology 39:869–873

Czigler I, Balázs L, Pató LG (2004) Visual change detection: event-related potentials are dependent on stimulus location in humans.Neurosci Lett 364:149–153

Czigler I, Weisz J, Winkler I (2006) ERPs and deviance detection:visual mismatch negativity to repeated visual stimuli. NeurosciLett 401:178–182

Czigler I, Weisz J, Winkler I (2007) Backward masking and visualmismatch negativity: Electrophysiological evidence for memory-based detection of deviant stimuli. Psychophysiology 44:610–619

Farrell DF, Leeman S, Ojemann A (2007) Study of the human visualcortex: direct cortical evoked potentials and stimulation. J ClinNeurophysiol 24:1–10

Fu S, Fan S, Chen L (2003) Event-related potentials reveal involuntaryprocessing of orientation changes in the visual modality. Psycho-physiology 40:770–775

Grice SJ, de Hahn M, Halit H, Johnson MH, Csibra G, Grant J,KarmiloV-Smith A (2003) ERP abnormalities of illusory contourperception in Williams Syndrome. Neuroreport 14:1773–1777

Hagen GF, Gatherwright JR, Lopez BA, Polich J (2006) P3a fromvisual stimuli: task diYculty eVects. Int J Psychophysiol 59:8–14

Halgren E, Dale AM, Mendola J, Chong CDR (2003) Cortical activa-tion to illusory shapes as measured with magnetoencephalogra-phy. Neuroimage 18:1001–1009

Heslenfeld DJ (2003) Visual mismatch negativity. In: Polich J (ed)Detection of change: event-related potential and fMRI Wndings.Kluwer, Dordrecht, pp 41–60

Horimoto R, Inagaki M, Yano T, Sata Y, Kaga M (2002) Mismatchnegativity of the color modality during selective attention task toauditory stimuli in children with mental retardation. Brain Dev24:703–709

Hruby T, Marsalek P (2003) Event-related potentials—the P3 Wave.Acta Neurobiol Exp 63:55–63

Inui K, Kakigi R (2006) Temporal analysis of the Xow from V1 to theextrastriate cortex in humans. J Neurophysiol 96:775–784

Kaiser J, Bühler M, Lutzenberger W (2004) Magnetoencephalographicgamma-band responses to illusory triangles in humans. Neuroim-age 23:551–560

Kanizsa G (1976) Subjective contours. Sci Am 234:48–52Kimura M, Murohashi H, Ji Katayama (2006) An ERP study of visual

change detection: EVects of magnitude of spatial frequencychanges on the change-related posterior positivity. Int J Psycho-physiol 62:14–23

Kraus N, Sharma A, McGee T, Carrell TD (1995) Neurophysiologicbases of speech discrimination. Ear Hear 16:19–37

Kremlacek J, Kuba M, Kubova Z, Langrova J (2006) Visual mismatchnegativity elicited by magnocellular system activation. VisionRes 46:485–490

Maekawa T, Taniwaki T, Kinukawa N, Kanba S, Goto Y, TobimatsuS (2005) Functional characterization of mismatch negativity to avisual stimulus. Clin Neurophysiol 116:2392–2402

Milne E, Scope A (2007) Are children with autistic spectrum disorderssusceptible to contour illusions? Br J Dev Psychol 26:91–102

Molholm S, Sehatpour P, Mehta AD, Shpaner M, Gomez-Ramirez M,Ortigue S, Dyke JP, Schwartz TH, Foxe JJ (2006) Audio-visualmultisensory integration in superior parietal lobule revealed byhuman intracranial recordings. J Neurophysiol 96:721–729

Näätänen R, Lehtokoski A, Lennes M, Cheour M (1997) Language-speciWc phoneme representations revealed by electric and mag-netic brain responses. Nature 385:432–434

123

Exp Brain Res (2009) 197:153–161 161

Näätänen R, Jacobsen T, Winkler I (2005) Memory-based or aVerentprocesses in mismatch negativity (MMN): a review of the evi-dence. Psychophysiology 42:25–32

Pazo-Alvarez P, Cadaveira F, Amenedo E (2003) MMN in the visualmodality: a review. Biol Psychol 63:199–236

Polich J (2003) Theoretical overview of P3a and P3b. In: Polich J (ed)Detection of change: event-related potential and fMRI Wndings.Kluwer, Dordrecht, pp 83–98

Polich J (2007) Updating P300: an integrative theory of P3a and P3b.Clin Neurophysiol 118:2128–2148

Schröger E (1997) On the detection of auditory deviations: a pre-atten-tive activation model. Psychophysiology 34:245–257

Schroger E, WolV C (1998) Attentional orienting and reorienting isindicated by human event-related brain potentials. Neuroreport9:3355–3358

Schroger E, Giard MH, WolV C (2000) Auditory distraction: event-re-lated potential and behavioral indices. Clin Neurophysiol111:1450–1460

Senkowski D, Rottger S, Grimm S, Foxe JJ, Herrmann CS (2005) Kan-izsa subjective Wgures capture visual spatial attention: evidencefrom electrophysiological and behavioral data. Neuropsychologia43:872–886

Shoji H, Ozaki H (2006) Topographic change in ERP due to discrimi-nation of geometric Wgures in the peripheral visual Weld. Int J Psy-chophysiol 62:115–121

Stagg C, Hindley P, Tales A, Butler S (2004) Visual mismatch nega-tivity: the detection of stimulus change. Neuroreport 15:659–663

Tales A, Newton P, Troscianko T, Butler S (1999) Mismatch negativ-ity in the visual modality. Neuroreport 10:3363–3367

Vogel EK, Luck SJ (2000) The visual N1 component as an index of adiscrimination process. Psychophysiology 37:190–203

Wallach H, Slaughter V (1988) The role of memory in perceiving sub-jective contours. Percept Psychophys 43:101–106

Winkler I, Reinikainen K, Naatanen R (1993) Event-related brainpotentials reXect traces of echoic memory in humans. PerceptPsychophys 53:443–449

123