■ Neurophysiological and BehaviouralCorrelates of Coherent MotionPerception in DyslexiaNaira A. Taroyan1*, Roderick I. Nicolson2 and David Buckley3

1Department of Psychology, Sheffield Hallam University, Sheffield, UK2Department of Psychology, Western Bank, University of Sheffield, Sheffield, UK3Academic Unit of Ophthalmology and Orthoptics, University of Sheffield, Sheffield, UK

Coherent motion perception was tested in nine adolescents with dyslexia and 10 controlparticipants matched for age and IQ using low contrast stimuli with three levels of coherence(10%, 25% and 40%). Event‐related potentials (ERPs) and behavioural performance data wereobtained. No significant between‐group differences were found in performance accuracy andresponse latencies of correct responses, for early (P1, N1, and P2) or late (P3) ERP peaks.However, attenuated early ERPs in the 10% coherent condition correlated significantly withlower performance accuracy (r=−.66) and with the magnitude of literacy deficit (r=−.46).Copyright © 2011 John Wiley & Sons, Ltd.

Keywords: dyslexia; coherent motion; event‐related potentials; behavioural response

INTRODUCTION

One of the central theoretical controversies in dyslexia research has been the potentialcausal role of magnocellular deficit, as reflected by impaired deficit in coherent visualmotion perception (e.g. Cornelissen et al., 1995; Eden et al., 1996; Kuba et al., 2001;Talcott et al., 2003; though see Vanni, Uusitalo, Kiesila, & Hari, 1997). Use of visual evokedpotentials to high contrast and low coherence level random‐dot kinetogram stimuli are apromising investigative tool, but studies have led to mixed results (e.g. Scheuerpflug et al.,2004; Schulte‐Korne, Bartling, & Deimel, 2004). It is possible that a reason for thebetween‐study discrepancies is the use of stimuli that might partially support local motion(e.g. Vanni et al., 1997) or parvocellular system (e.g. Scheuerpflug et al., 2004) processing.This study investigated global coherent motion perception in dyslexia by means ofrecording ERPs to random‐dot kinetogram stimuli that combined low contrast and lowcoherence levels, thereby limiting processing to the magnocellular system.

METHODS

Participants

Nine dyslexic adolescents (three women) and 10 control adolescents (three women) tookpart. All were right‐handed, with normal or corrected to normal vision, with no

*Correspondence to: Naira A. Taroyan, Department of Psychology, Sheffield Hallam University, Sheffield S102BP, UK. E‐mail: N.A.Taroyan@shu.ac.uk

Copyright © 2011 John Wiley & Sons, Ltd. DYSLEXIA 17: 282–290 (2011)

DYSLEXIAPublished online in Wiley Online Library(wileyonlinelibrary.com). DOI: 10.1002/dys.435

overlapping ADHD symptoms and no history of brain injuries or neurological problems.Mean values of age, IQ (Wechsler, 1991), reading age (RA) and spelling age (SA)(Wechsler, 1993) for both groups are shown in Table 1. Criteria for inclusion in thedyslexic group were a discrepancy of at least 18months between an individual’s readingand spelling age and his/her chronological age together with an IQ> 90,1 which arestandard UK criteria (e.g. Seymour, 1986; Thomson, 1984). There were no significantdifferences between the dyslexic and control groups on IQ scores (one‐factor ANOVA,p> .1), but as expected, dyslexics had significantly lower RA (p< .0002) and SA(p< .0001) scores compared with controls.

Stimuli

Random dot stimuli were designed in Matlab using Psychophysics Toolbox (Brainard,1997; Pelli, 1997) and were displayed on a 20 inch PC monitor (1028 × 768 pixelresolution, 75Hz refresh rate). The stimuli were low contrast (Michelson 5%) andconsisted of black dots displayed with an average density of 4 dots/deg2 on a light greybackground, which subtended a 30 deg × 20 deg angular size at a viewing distance of60 cm. Each 2 × 2 pixel dot subtended 0.06 deg2 size, moving at a speed of 5 deg/s with alifetime of 100ms.

Design/Paradigm

There were three coherent conditions with 10%, 25% or 40% of the dots moving upwards,whereas the remaining dots moving in pseudo‐random directions with each frame. Therewas also one incoherent (control) condition where all dots moved in pseudo‐randomdirections within each frame. Equal numbers of coherent trials (40 for each of the threeconditions) and incoherent trials (120) were presented in quasi random order. Each trialstarted with a small fixation cross displayed in the centre of the screen (1000ms), followedby the stimulus (either coherent or incoherent motion, 500ms) also containing the fixationcross, followed by a blank screen (1500ms). The participants pressed one button on theresponse box when they saw coherent motion and the other button if they perceivedrandom motion.

Data Acquisition

The electroencephalogram (EEG) was recorded by a 128 channel (plus a Cz reference atthe vertex) Geodesic Sensor Net (Tucker, 1993) of Electrical Geodesics, Inc. (EGI,Eugene, OR, USA), amplified by a factor of ×1000 by EGI high‐input impedance amplifier(200MΩ, Net Amps) with 0.1‐100Hz bandpass‐filter, and digitised with a 12‐bit A/Dconverter at a sampling frequency of 250Hz. The responses and response latencies ofcorrect responses were collected (simultaneously with the EEG data) by a PowerMacintosh OS X (10.2.8) and stored for off‐line analysis.

1One dyslexic participant had improved his reading age and spelling age scores because the time of diagno-sis, however, his behavioural and event‐related potentials data were characteristic of dyslexic groupaverages. The statistical analysis performed without this subject did not change the significance values ofneither event‐related potentials, nor behavioural results or the group effect size.

ERP Correlates of Motion Perception in Dyslexia 283

Copyright © 2011 John Wiley & Sons, Ltd. DYSLEXIA 17: 282–290 (2011)

Data Analysis

Mean response latencies of correct responses and the overall accuracy (percentage ofcorrect responses) were determined for each participant and condition. The EEG datawere analysed using NetStation 4.0 EGI software (Electrical Geodesics, Inc., Eugene, OR,USA) (for details, see Taroyan & Nicolson, 2009; Taroyan, Nicolson, & Fawcett, 2007).The group average ERPs in each condition were computed separately for the dyslexicand control participants. Early P1 (~140ms from stimulus onset), N1 (~190ms), and P2(~270ms) ERP components were identified in left and right occipital (lower insets inFigure 1) regions, whereas later P3 peak (~560ms) was identified in the left and rightparietal (upper insets in Figure 1) sites. This late positive peak is typically found at parietallocations within a broad latency window of 300–600ms (e.g. Dien, Spencer, & Donchin,2004; Polich & Donchin, 1988).

N1

Group average ERPs in Coherent 40% condition

ControlsDyslexic

10 µV

P1 P2

N1

P3P3

1000ms

P1 P2

Figure 1. Group average event‐related potentials (ERPs) for control ( ) and dyslexic ( ) participants incoherent 40% condition at all electrode sites. The larger scale waveforms in the top and bottomcorners show the representative ERPs and characteristic peaks from parietal (upper insets ofchannels) and occipital (lower insets of channels) regions. The vertical lines on the waveforms

(arrows on larger scale ERPs) indicate the stimulus onset at 0ms.

Table 1. Psychometric mean data (ranges in parentheses)

Controls (N= 10) Dyslexic (N= 9)

Age 16.6 (15.4–19.3) 17.1 (15.6–17.8)IQ 119.7 (103–132) 111.5 (99–124)Reading age 17+ (maximum) 13.0 (9.3–17+)Spelling age 17+ (maximum) 11.5 (8.6–17+)

284 N. A. Taroyan et al.

Copyright © 2011 John Wiley & Sons, Ltd. DYSLEXIA 17: 282–290 (2011)

Statistical Analysis

Preliminary data analysis revealed that one dyslexic participant was an outlier in some ofthe behavioural and ERP measures. This making many more errors in the 10% coherentmotion condition (more than three standard deviations (SD) larger than the mean of therest of dyslexic participants). The latency of N1 and P2 ERP components in all conditionsfor this participant were also more than three SD longer than the mean values for the restof the group. Therefore, the data of this participant were replaced with the next highestvalue from the group (Field, 2009) in further statistical analysis.

Two‐factor mixed measures ANOVA were undertaken on the behavioural data, withone between‐subjects factor group (dyslexic versus control) and one four‐level within‐subjects factor condition (three coherent and one incoherent).

The amplitudes of the peaks in the ERPs of individual subjects were found in the timewindows defined by the N1, P1 and P3 peaks in corresponding channel groups (as detailedearlier and in Figure 1) average ERPs and automatically measured relative to the pre‐stimu-lus baseline. The latency of these peaks was computed relative to the stimulus onset.Three‐factor mixed measures ANOVAs were undertaken on these ERP amplitude andlatency data, with the third factor being Hemisphere (left and right).

Additional correlational analyses were performed in order to establish whether the indicesof the brain activation correlatedwith the behavioural performance, and with the magnitude ofthe ‘discrepancy’ between reading age and chronological age, acting as an indicator for‘strength’ of dyslexia. Two measures of discrepancy were used, spelling age discrepancy andreading age discrepancy. Only the results for the most difficult condition (10% coherence) thatreached significance (p< .05) are reported in the succeeding text.

RESULTS

Behavioural Data

There was a significant main effect of condition for both dependent variables [F(3,51) = 17.7,p< .0001; F(3,51) = 7.1, p< .001 for latencies and errors, respectively]. Response latencieswere longer in the incoherent and coherent 10 % conditions compared with the coherent40% and coherent 25% conditions across both groups (Figure 2a). More errors were madein the coherent 10% and incoherent conditions compared with the 40% and 25% coherentconditions (Figure 2b).

A signal detection analysis was undertaken on the accuracy data The two‐factorANOVA did not reveal any significant between‐group differences for the sensitivity(dprime; [F(1,17) = 2.9]) or the response bias (beta; [F(1,17) = 0.3]) values (seeFigure 2c–d).

Event‐Related Potential Data

According to the results of ANOVA the amplitude of P1 and P2 was larger in the coherent10% condition compared with the other conditions [F(3,51) = 3, p< .05; F(3,51) = 3.1,p< .05, respectively] for the controls but not for the dyslexics, as can be seen in Figure 3a.

It can also be observed in Figures 1, 3b and topographic maps of Figure 4 that the P3amplitude was larger in the coherent 40% condition compared with the other conditions[F(3,51) = 5.45, p< .01], and it was also larger in the right compared with the left

ERP Correlates of Motion Perception in Dyslexia 285

Copyright © 2011 John Wiley & Sons, Ltd. DYSLEXIA 17: 282–290 (2011)

hemisphere across both groups [F(1,17) = 7.3, p< .05]. There were no significant between‐group ERP differences.

The brain‐behaviour correlational analyses revealed that the P1 amplitude in the righthemisphere in the coherent 10% condition was significantly negatively correlated withthe spelling deficit (r=−.455, p< .05), and the sensitivity of the performance in 10%coherent motion condition significantly positively correlated with the P1 amplitude(r= .659, p< .01).

DISCUSSION

The behavioural performance of our participants was relatively good in both groups with only10%–20% of errors made in perceptually difficult coherent 10% and incoherent conditions,which may reflect the fact that we used coherent/incoherent judgement task as opposed tomotion direction detection (Cornelissen & Hansen, 1998; Cornelissen et al., 1995).

The increased P3 amplitude in the highly (40%) coherent motion condition in the rightcompared with the left hemisphere across both groups replicates the results fromprevious studies showing increased brain activation with higher coherence levels ofmotion (Patzwahl & Zanker, 2000; Schulte‐Korne et al., 2004), as well as higher sensitivityof the right hemisphere in coherent motion processing (Niedeggen & Wist, 1999). Theincreased P1 and P2 amplitude and longer response latencies in the 10% compared with40% and 25% coherent conditions may be a result of increased attentional efforts andmore brain activation required for processing of this perceptually difficult stimulus.Although no significant between‐group ERP differences were found in our sample ofparticipants, the brain‐behaviour correlational analysis findings provide suggestive evidencethat literacy problems and decreased performance sensitivity in perceptually difficult 10%coherent motion condition are associated with attenuated brain activation.

500

550

600

650

700

750

40 25 10 0

Res

po

nse

late

ncy

(m

s)

Coherence level (%)

0

1

2

3

4

5

6

40 25 10

Sen

siti

vity

(d

pri

me)

Coherence level (%)

0

5

10

15

20

25

30

40 25 10 0

Inco

rrec

t res

po

nse

s (%

)

Coherence level (%)

b

a

-1.4

-1.0

-0.6

-0.2

0.2

0.6

40 25 10

Bia

s (b

eta)

Coherence level (%)

d

c

Figure 2. Behavioural data for control and dyslexic groups (mean ± SE) for (a) response latencies, (b)performance, (c) dprime and (d) bias.

286 N. A. Taroyan et al.

Copyright © 2011 John Wiley & Sons, Ltd. DYSLEXIA 17: 282–290 (2011)

Future research with larger numbers of dyslexic and control participants would beneeded to identify an accurate indicator of the incidence of the magnocellular deficits inindividuals with dyslexia, and indeed to indicate whether the incidence is higher than in

a

Figure 3. Event‐related potential waveforms for control and dyslexic groups in occipital (a) andparietal (b) left and right areas (averages of channel group insets shown in Figure 1) separately for

each motion condition.

ERP Correlates of Motion Perception in Dyslexia 287

Copyright © 2011 John Wiley & Sons, Ltd. DYSLEXIA 17: 282–290 (2011)

the control population. Larger numbers of participants should also shed light on the issueof whether our one ‘outlier’, who performed extremely poorly in the 10% condition, doesin fact represent a significant sub‐type of dyslexia with a marked magnocellular problem.

In conclusion, the present study was designed to probe whether visual sensory deficitsin dyslexia are revealed in ERP data when the opportunity for parvocellular processing is

b

Figure 3. Continued

288 N. A. Taroyan et al.

Copyright © 2011 John Wiley & Sons, Ltd. DYSLEXIA 17: 282–290 (2011)

eliminated. For the participants with dyslexia, there were few clear differences from thecontrols, though there were interesting correlations between magnitude of literacy deficitand amplitude of the early ERP components in the most difficult, low coherence, condition.

ACKNOWLEDGEMENTS

This work was supported by a CASE Collaborative Research Studentship awarded by theEconomic and Social Research Council (ESRC) to the University of Sheffield. We wouldlike to thank our participants and their parents for cooperation and our anonymousreviewers for their valuable contributions to our manuscript.

REFERENCES

Brainard, D. H. (1997). The psychophysics toolbox. Spatial Vision, 10, 433–436.Cornelissen, P., Richardson, A., Mason, A., Fowler, S., & Stein, J. (1995). Contrast sensitivity andcoherent motion detection measured at photopic luminance levels in dyslexics and controls. VisionResearch, 35, 1483–1494.Cornelissen, P. L., & Hansen, P. C. (1998). Motion detection, letter position encoding, and singleword reading. Annals of Dyslexia, 48, 155–188.Dien, J., Spencer, K. M., & Donchin, E. (2004) Parsing the late positive complex: Mental chronometryand the ERP components that inhabit the neighborhood of P300. Psychophysiology, 41, 665–678.Eden, G. F., Vanmeter, J. W., Rumsey, J. M., Maisog, J. M., Woods, R. P., & Zeffiro, T. A. (1996).Abnormal processing of visual motion in dyslexia revealed by functional brain imaging. Nature,382, 66–69.Field, A. P. (2009). Discovering statistics using SPSS. London: Sage Publications Ltd.Kuba M., Szanyi, J., Gayer, D., Kremlacek, J., & Kubová Z. (2001). Electrophysiological testing ofdyslexia. Acta Medica, 44(4), 131–134.

Controls

560 ms

25%40% 10% Incoherent

Dyslexics

3µV

-3µV

Figure 4. Topographic event‐related potential maps for P3 peak at 560ms in all conditions for con-trol and dyslexic groups. The black spots represent 128 channels, as in Figure 1.

ERP Correlates of Motion Perception in Dyslexia 289

Copyright © 2011 John Wiley & Sons, Ltd. DYSLEXIA 17: 282–290 (2011)

Niedeggen, M., & Wist, E. R. (1999). Characteristics of visual evoked potentials generated by motioncoherence onset. Cognitive Brain Research, 8, 95–105.Patzwahl, D. R., & Zanker, J. M. (2000). Mechanisms of human motion perception: Combiningevidence from evoked potentials, behavioural performance and computational modelling. EuropeanJournal of Neuroscience, 12, 273–282.Pelli, D. G. (1997). The VideoToolbox software for visual psychophysics: Transforming numbers intomovies. Spatial Vision, 10, 437–442.Polich, J. & Donchin, E. (1988) P300 and the word frequency event. Electroencephalography andClinical Neurophysiology, 70, 33–45.Scheuerpflug, P., Plume, E., Vetter, V., Schulte‐Koerne, G., Deimel, W., Bartling, J., Remschmidt, H.,& Warnke, A. (2004). Visual information processing in dyslexic children. Clinical Neurophysiology, 115,90–96.Schulte‐Korne, G., Bartling, J., & Deimel, W. (2004). Visual evoked potential elicited by coherentymoving dots in dyslexic children. Neuroscience Letters, 357, 207–210.Seymour, P. H. K. (1986). Cognitive analysis of dyslexia. London: Routledge & Kegan Paul.Thomson, M. E. (1984). Developmental dyslexia: Its nature, assessment and remediation. London:Edward Arnold.Talcott, J. B., Gram, A., Van Ingelghem, M., Witton, C., Stein, J. F., & Toennessen, F. E. (2003).Impaired sensitivity to dynamic stimuli in poor readers of a regular orthography. Brain and Language,87, 259–266.Taroyan, N. A., & Nicolson, R. I. (2009). Reading words and pseudowords in dyslexia: ERP and be-havioural tests in English speaking adolescents. International Journal of Psychophysiology, 74, 199–208.Taroyan, N. A., Nicolson, R. I., & Fawcett, A. J. (2007). Behavioural and neurophysiological correlatesof dyslexia in the continuous performance task. Clinical Neurophysiology, 118, 845–855.Tucker, D. (1993). Spatial sampling of head electrical fields: The geodesic sensor net. Electroenceph-alography and Clinical Neurophysiology, 87, 154–163.Vanni, S., Uusitalo, M. A., Kiesila, P., & Hari, R. (1997). Visual motion activates V5 in dyslexics.Neuroreport, 8, 1939–1942.Wechsler, D. (1991). Wechsler intelligence scale for children (3rd ed.). Sidcup UK: The PsychologicalCorporation, Europe.Wechsler, D. (1993). Wechsler objective reading dimension. London: The Psychological Corporation.

290 N. A. Taroyan et al.

Copyright © 2011 John Wiley & Sons, Ltd. DYSLEXIA 17: 282–290 (2011)