+ All Categories
Home > Documents > Early electro-cortical correlates of inspection time task performance

Early electro-cortical correlates of inspection time task performance

Date post: 21-Oct-2016
Category:
Upload: david-hill
View: 213 times
Download: 0 times
Share this document with a friend
8
Early electro-cortical correlates of inspection time task performance David Hill a , Christopher W.N. Saville a, b , Siobhan Kiely a , Mark V. Roberts a , Stephan G. Boehm a , Corinna Haenschel a, c , Christoph Klein a, b, a School of Psychology, Bangor University, United Kingdom b Department of Child and Adolescent Psychiatry and Psychotherapy, University of Freiburg, Germany c Department of Psychology, City University, London, United Kingdom article info abstract Article history: Received 26 January 2011 Received in revised form 7 June 2011 Accepted 9 June 2011 Available online 2 July 2011 The concept of general intelligence (g) summarizes the well established nding that scores on separate cognitive tasks are positively correlated, indicating a trait common to many aspects of information processing. Inspection time is a well-established correlate of IQ, where those of a higher IQ can correctly identify a briey presented stimulus with a greater level of accuracy than those of a lower IQ. This study used two age-matched samples, selected on the basis of their scores on Raven's Advanced Progressive Matrices from the undergraduate population of Bangor University. In order to address the confound of inspection time and IQ of previous ITT-ERP research, each participant of the presented study performed an IT task with the same ve levels of stimulus duration while undergoing 64-channel EEG recording. The high IQ group made signicantly fewer errors at each level of stimulus duration and exhibited a signicantly larger N1 response. N1 latency and other ERP components did not distinguish the two IQ groups. Given the specicity of ERP group differences to the N1, the results of the present study suggest that the link between IT performance and g is attributable to individual differences in directing attention to a spatial region. © 2011 Elsevier Inc. All rights reserved. Keywords: Inspection time Raven's advanced progressive matrices ERP N1 1. Introduction In the attempt to elucidate the nature of g, intelligence researchers have sought to establish cognitive correlates of g using more elementary tasks, such as those assessing processing speed, working memory or sustained attention (Schweizer, 2005). After more than 50 years of research, processing speed can be considered an established cognitive correlate of psychometric intelligence (Sheppard & Vernon, 2008). Processing speed is typically assessed with cognitively simple tasks such as the Inspection Time Task (ITT; Deary & Stough, 1996). During this task, participants have to identify simple, but extremely briey presented, masked patterns. The inspection time (IT) a participant needs to produce a correct response varies greatly between individuals (Vickers, Nettelbeck, & Willson, 1972) and these individual differences have been consistently found to be correlated with individual differences in tests of psychometric intelligence (Grudnik & Kranzler, 2001). The high correlation between performance in this simple speeded task on the one side and performance in complex non-speeded problem solving tasks such as Raven's Progressive Matrices is doubtlessly among the most intriguing relationships that research into individual differ- ences has established. It therefore calls both for a convincing theoretical explanation and the elucidation of its neural correlates. Exploratory factor analysis performed by Burns and Nettelbeck (2003) has shown that IT ts into the hierarchy of human cognitive abilities through a general speediness factor, with IT being most closely related to visualization speed (O'Connor & Burns, 2003). However as to whether this reects speed of post sensory or sensory processes or a combination of the two remains unclear. On the one side, Intelligence 39 (2011) 370377 Corresponding author at: School of Psychology, Bangor University, The Brigantia Building, Penrallt Road, Bangor, Gwynedd LL57 2AS, Wales, UK. Tel.: +44 1248 38 8351; fax: +44 1248 38 2499. E-mail address: [email protected] (C. Klein). 0160-2896/$ see front matter © 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.intell.2011.06.005 Contents lists available at ScienceDirect Intelligence
Transcript
Page 1: Early electro-cortical correlates of inspection time task performance

Intelligence 39 (2011) 370–377

Contents lists available at ScienceDirect

Intelligence

Early electro-cortical correlates of inspection time task performance

David Hill a, Christopher W.N. Saville a,b, Siobhan Kiely a, Mark V. Roberts a, Stephan G. Boehm a,Corinna Haenschel a,c, Christoph Klein a,b,⁎a School of Psychology, Bangor University, United Kingdomb Department of Child and Adolescent Psychiatry and Psychotherapy, University of Freiburg, Germanyc Department of Psychology, City University, London, United Kingdom

a r t i c l e i n f o

⁎ Corresponding author at: School of Psychology, BBrigantia Building, Penrallt Road, Bangor, GwyneddTel.: +44 1248 38 8351; fax: +44 1248 38 2499.

E-mail address: [email protected] (C. Klein).

0160-2896/$ – see front matter © 2011 Elsevier Inc.doi:10.1016/j.intell.2011.06.005

a b s t r a c t

Article history:Received 26 January 2011Received in revised form 7 June 2011Accepted 9 June 2011Available online 2 July 2011

The concept of general intelligence (g) summarizes the well established finding that scores onseparate cognitive tasks are positively correlated, indicating a trait common to many aspects ofinformation processing. Inspection time is a well-established correlate of IQ, where those of ahigher IQ can correctly identify a briefly presented stimulus with a greater level of accuracythan those of a lower IQ. This study used two age-matched samples, selected on the basis oftheir scores on Raven's Advanced Progressive Matrices from the undergraduate population ofBangor University. In order to address the confound of inspection time and IQ of previousITT-ERP research, each participant of the presented study performed an IT task with the samefive levels of stimulus duration while undergoing 64-channel EEG recording. The high IQ groupmade significantly fewer errors at each level of stimulus duration and exhibited a significantlylarger N1 response. N1 latency and other ERP components did not distinguish the two IQgroups. Given the specificity of ERP group differences to the N1, the results of the present studysuggest that the link between IT performance and g is attributable to individual differences indirecting attention to a spatial region.

© 2011 Elsevier Inc. All rights reserved.

Keywords:Inspection timeRaven's advanced progressive matricesERPN1

1. Introduction

In the attempt to elucidate the nature of g, intelligenceresearchers have sought to establish cognitive correlates of gusing more elementary tasks, such as those assessingprocessing speed, working memory or sustained attention(Schweizer, 2005). After more than 50 years of research,processing speed can be considered an established cognitivecorrelate of psychometric intelligence (Sheppard & Vernon,2008). Processing speed is typically assessed with cognitivelysimple tasks such as the Inspection Time Task (ITT; Deary &Stough, 1996). During this task, participants have to identifysimple, but extremely briefly presented, masked patterns.The inspection time (IT) a participant needs to produce a

angor University, TheLL57 2AS, Wales, UK

All rights reserved.

.

correct response varies greatly between individuals (Vickers,Nettelbeck, & Willson, 1972) and these individual differenceshave been consistently found to be correlated with individualdifferences in tests of psychometric intelligence (Grudnik &Kranzler, 2001). The high correlation between performancein this simple speeded task on the one side and performancein complex non-speeded problem solving tasks such asRaven's Progressive Matrices is doubtlessly among the mostintriguing relationships that research into individual differ-ences has established. It therefore calls both for a convincingtheoretical explanation and the elucidation of its neuralcorrelates.

Exploratory factor analysis performed by Burns andNettelbeck (2003) has shown that IT fits into the hierarchyof human cognitive abilities through a general speedinessfactor, with IT being most closely related to visualizationspeed (O'Connor & Burns, 2003). However as to whether thisreflects speed of post sensory or sensory processes or acombination of the two remains unclear. On the one side,

Page 2: Early electro-cortical correlates of inspection time task performance

371D. Hill et al. / Intelligence 39 (2011) 370–377

individual differences in IT performance were originallyhypothesized to be due to individual differences in the timerequired to make a single observation, this being all that isrequired to make a correct discrimination in the IT task(Vickers & Smith, 1986). More recent accounts of theinspection time task performance stress that IT is a measureof the quality of stimulus representation and memorycapacity (Vickers, Pietsch, & Hemingway, 1995), and arethought to occur at a post sensory level. Indeed, a number ofworks have examined the role attention may play in the ITtask (Bors, Stokes, Forrin, & Hodder, 1999; Hutton, Wilding, &Hudson, 1997). Other works theorize that individual differ-ences in inspection time may be due to sensory systemsalone, whereby the correlation between g and IT is due to ITbeing linked to neuronal activity that is common to bothcognitive and sensory processes (White, 1993, 1996).

However, evidence from Caryl (1994), on the other side,whomanipulated the interval between cue and stimulus in theITT, has suggested a role for “higher level” cognitive abilitiessuch as top down deployment of attention. “Lower level” and“higher level” explanatory approaches are not necessarilyirreconcilable. For example, it has been suggested that proces-sing speed also affects the quality of the processing of complexmaterial because of the capacity and temporal limitations ofstoring information in short-term/working memory, anotherestablished cognitive correlate of psychometric intelligence(Colom, Abad, Quiroga, Shih, & Flores-Mendoza, 2008; forcomments on this proposition, see Necka, 1992). The estab-lished positive relationships between mental speed in simpletasks and accuracy in complex tasks on the one side, andaccuracy in complex tasks and success in solvingRaven'smatrixproblems on the other side support the aforementionedassumption of a speed–accuracy transition (Schweizer, 1996).And indeed, task analyses of Raven's matrices have pointed tothe importance of the ability to identify abstract relationshipsand to dynamicallymanage a large set of problem-solving goalsin working memory when solving this task (Carpenter, Just, &Shell, 1990).

Regarding the neural correlates ofmental speed, researchershave exploited the excellent temporal resolution of eventrelated potentials (ERP) to investigate the neurophysiology ofITT performance. Converging lines of research have shown thatsubjects performing relativelywell in the ITT showa fasterpeak-to-peak transition from N1 to P2 than participants performingrelatively poorly in this task (Caryl, 1994; Zhang, Caryl, & Deary,1989a). The latency range at which the shift from N1 to P2occurs corresponds to a stage at which information has movedfrom the striate cortex to the extra-striate cortex (Mangun,1995). It is in this region where top-down factors includingattention are thought to be able to operate on information tofacilitate or attenuate processing (Mangun, 1995). That thedifference between the waveforms of those with high ascompared to low ITT performance occurs early, is, atfirst glance,compatible with the aforementioned assumption of the ITTtaxing “lower level” processes (Vickers & Smith, 1986).

However, while the N1–P2 slope significantly correlateswith IQ (Zhang, Caryl, & Deary, 1989b), this measure suffersfrom the ambiguity of all peak-to-peak measures: thedifference is between two components, and it is unclearwhat such a difference represents. Accordingly, their corre-lations with ITT and IQ performance should be investigated

individually. This holds all the more as the N1 and P2 reflectrelated but distinct processes (Luck & Hillyard, 1994; Vogel &Luck, 2000). N1 and P2 belong to an early ERP complex thattypically includes another component: the P1.

The visual P1 is a positive component peaking 60–100 msafter stimulus onset over lateral occipital regions (Clark &Hillyard, 1996). This component is alsomodulated by attention,with greater amplitudes, faster reaction times and greateraccuracy, in response to stimuli appearing at attended locationscompared with unattended locations (Mangun & Hillyard,1991). The P1 amplitude is, furthermore, thought to reflect thesuppression of information outside of attended space (Hillyard,Vogel, & Luck, 1998), which results in greater signal fromattended regions due to less competing information.

The visual N1 is a negative component that peaks 150 to200 ms after stimulus presentation over lateral occipitalregions (Hopf, Vogel, Woodman, Heinze, & Lucky, 2002).Attention directed to a region of space results in an increase inthe amplitude of the P1 and N1. While this was initiallythought to reflect a process common to both of thesecomponents, subsequent research has found that the P1 andthe N1 reflect different aspects of information processing. TheN1 component is elicited in tasks where a subject must firstattend to a region of space and then discriminate betweentarget and non-target stimuli (Ritter, Simson, Vaughn, &Macht, 1982). This discrimination effect has been found fortargets that differ in color or structural form but is not presentin tasks that require only the detection of a stimulus (Vogel &Luck, 2000). The N1 component therefore appears to reflectan enhancement of features at an attended location.

The visual P2 component, finally, is a positivewave formationpeaking 150–275 ms after stimulus onset over anterior andcentral electrode sites (Luck & Hillyard, 1994). An enhanced P2can be seen in visual search taskswhereparticipants are requiredto locate a target defined by simple features, be they color(Hillyard&Munte, 1984)or orientationand size (Luck&Hillyard,1994). This enhancement of the P2 is not present in visual searchdisplays that do not contain the target feature (Luck & Hillyard,1994). The P2 has been thought to reflect a process whereby thefeatures of a stimulus that define it as a target are amplified tofacilitate its detection (Luck & Hillyard, 1994).

To summarize, while all three of these components aremodulated by attention, they represent different aspects ofattention. The P1 appears to represent the suppression ofunattended locations, which facilitates target detection(Hillyard et al., 1998), while the N1 represents task relevantfeatures at a location being selectively enhanced to facilitatediscrimination (Vogel & Luck, 2000). The P2 by contrastappears to represent the facilitation of the detection ofspecific features (Luck & Hillyard, 1994).

That ITT performance is linked bothwith g (Deary & Stough,1996) andwith the latencies andamplitudesof early perceptualERPs (Burns, Nettelbeck, & Cooper, 2000) suggests thatintelligence and the speed of those processes that are indexedby the P1, N1 or P2 should be correlated aswell. Unfortunately,previous research into the relationship between IQ and such“early”ERPs using the ITT, experimentally confounded stimuluspresentation duration with participants' IQ. In order to equatethe IQ groups in terms of ITT performance, longer stimulusdurations were presented to the less intelligent participants(Caryl, 1994; Zhang et al., 1989b).

Page 3: Early electro-cortical correlates of inspection time task performance

372 D. Hill et al. / Intelligence 39 (2011) 370–377

Based on these considerations, we abandoned the previousexperimental confounding of IQ group and a potentially crucialexperimental setting by parametrically varying stimuluspresentation durations in order to pursue the followingresearch goals. First, we wanted to compare group effect sizesof the separately analyzed early ERP components in order todisentangle which of the different processes is most closelyrelated to IQ. And second, we combined the assessment ofwithin-session retest reliabilities of the involvedmeasureswitha test of the hypothesis of a relative sustained attention "deficit"in low IQ participants by comparing ITT performance and ERPmeasures at the beginning and at the end of an extendedexperimental ERP session.

2. Methods

2.1. Participants

Forty undergraduate students from Bangor University wereselected on the basis of their scores on Raven's AdvancedProgressiveMatrices (RAPM; Raven, Court, & Raven, 1993). Thescreening took place as part of a module on personality andindividual differences, during which 300 participants werescreened in groups of 20 or 45. RAPM set two was conductedunder exam conditions with the first problemworked throughasanexample andwith 50 minallotted to the completionof theremaining 35 items. One group was selected from the 15th–25th percentile of RAPMdistribution. This “low IQ” group had araw RAPM score of 14.74±1.59 (M±SD; range 13–16). Asecond group was selected from the 50th–75th percentile; this“high IQ” group had a raw RAPM score of 21.05±1.36 (range20–25). Quotation marks indicate that “low” and “high” areused to denote the relative position of the two IQ groups and donot indicate their absolute position in the population. Themeanage of participants in the “high IQ” groupwas 21.43±3.7 years,47.6% of which were male. The “low IQ” group had a mean ageof 21.74±3.28 years, 22.2% of which were male (groupdifferences: age: t(38)=.277, p=.78; gender: χ2(38)=1.61,p=.32). Participants had previously given consent to becontacted for this follow-up study as part of the IQ screening.Participants who took part in this study were informed of theirright to withdraw at any point. All participants had normal orcorrected to normal vision, were right-handed and free of anyneurological or psychiatric conditions. Participants were paid£45 for taking part in this study.

2.2. Apparatus and physiological recordings

All stimuli appeared on a 19″ cathode ray tube monitorwith a refresh rate of 85.0 Hz using E-Prime software(Psychology Software Tools, Sharpsburg). The experimenttook place within a sound attenuated Faraday cage housing amonitor with an electrically shielded power unit. The EEGwas recorded from 64 positions of the international 10–10system (American Electroencephalographic Society, 1991)along with three additional infra-orbital and nasion elec-trodes, positioned at IO1, IO2 and Nz using 8 mm adhesivepads. The ring electrodes were held in place using an elasticcap (Easy Cap; FMS, Munich) with the conduction facilitatedwith the Abralyt-light (FMS, Munich) electrode gel. TwoBrainAmps DC amplifiers with a resolution of 0.1 μV (Brain

Products, Munich) were used to record the EEG at a samplingrate of 500 Hz using an analog low pass filter of 250 Hz (DC—250 Hz band-pass). Impedances at all electrode sites weremaintained below 5 kΩ. During recording the electrodeswere referenced to Cz and the ground electrode was placed atAFz.

2.3. Stimuli and procedures

Each participantwas tested individually. All participants satapproximately 60 cm in front of the screen andwere presentedwith a black horizontal line 18 mmacross atfixation on awhitebackground. After 2000 ms, the screen went blank for 300 msfollowed by the Π-stimulus at the same spatial location. Thisstimulusmeasured18 mmacross, the short legwas13 mmandthe long leg 23 mm. A backward visual mask, 18 mm across,with each leg 30 mm in length appeared for the durationrequired to make the duration of stimulus plus maskpresentation equal 1000 ms (see Fig. 1). Stimuli with long orshort legs were presented in random order. Participantsundertook 20 supervised practice trials of the 58.83 mscondition to ensure that they understood the task. Participantstook part in two ITT blocks that were separated by a series ofworking memory tasks (0-back, 1-back) which took about30 min andwere conducted as part of a separate study thatwillbe presented elsewhere. Participants were presented with 100trials for each of the five stimulus durations; the left leg wouldbe the shorter in 50 trials (fixed pseudo-random order). Withthe two ITT blocks totaled this made 1000 trials. Theparticipants were instructed to respond to the shortest leg bypressing the “A” key for left and the “L” key for the right.Accuracy, but not speed, was stressed by the task instruction.After the experiment the participants were thanked anddebriefed. The study was approved by the departmental ethicscommittee, and written informed consent was obtained fromall participants in accordance with the Helsinki declaration.

2.4. Processing of EEG data and statistical analyses

Primary data processing was accomplished off-line withBrain Vision Analyser (Version 2.0, Brain Products, Munich).EEGdatawere re-referenced to the common average reference,followed by a raw data inspection to identify and excludestretches of EEG activity, the upper range was set at 1500 μVover a 500 ms epoch and the lower was set to 0.5 μV over100 ms. This was followed by an Infomax IndependentComponent Analysis on an epoch of 6–240 s (ICA; Makeig etal., 1999). Components related to eye movements wereidentified by their topography and association with triggers,which indicated the presence of a saccade or a blink, and wereexcluded from back-projection. Trials with correct responseswere separated into epochs of 2000 ms, starting 1000 msbefore stimulus onset and concluding 1000 ms post stimulus.The data were then visually examined for the presence of anyother extraneous components, which were also removed,followed by a baseline correction 1000–500 ms before stimuluspresentation. The data were then filtered leaving only thefrequencies between 1.0 and 12.0 Hz. All artifact-free epochswith correct responses were averaged into ERPs for both IQgroups. ERP components were determined by their topogra-phies and peak latencies. The P1 was identified as the first

Page 4: Early electro-cortical correlates of inspection time task performance

Cue

Gap

Stimulus

Mask

Response

Fig. 1. The Inspection Time Task shown with left leg shorter.

373D. Hill et al. / Intelligence 39 (2011) 370–377

positive deflection occurring 70–124 ms post stimulus locatedin the left hemisphere at electrodes PO7 and P5. Theseelectrodes were averaged for the analysis of the P1. Theaggregated electrode channels for the P1 correlated in excess of.70. The N1 was defined as the first negative deflectionoccurring 125–190 ms post stimulus located at P3 electrode.Due to thebilateral nature of theN1wave formation (seeFig. 4),four electrodes from each hemisphere were averaged for theanalysis of the N1 (P3, PO3, P5, CP3, CP4, CP6, P2 and P4). Allcorrelations for the aggregated ipsilateral electrode channelswere in excess of .82. The P2 was identified as the secondpositive deflection occurring 170–260 ms post-stimulus locat-ed centrally over the electrodes PZ, PO3 and PO4. Thesechannels were correlated in excess of .65 and were thusaggregated for the analysis of the P2. The ERP components P1,N1 and P2 were separately analyzed using mixed factorialANOVAs, including the between-subjects factor GROUP (highvs low IQ) and thewithin-subjects factors IT DURATION (11.80,23.60, 35.30, 47.06, 58.83 ms) and HEMISPHERE (P3, P5, CP3,PO3: left hemisphere; P4, P2, CP4, CP6: right hemisphere). Acomputer error caused a number of the participants' data to be

Percentage correct

Duration of Π-figure in ms

High IQ

Low IQ

Fig. 2. ITT performance in the high- and low-IQ groups. This shows that thehigh IQ group had a significantly greater proportion of correct responses thanthe low IQ group at the each level of stimulus presentation. Error bars show±1 standard error of the mean.

Table 1Inspection Time Task performance (in percentage correct) in the high- and low-IQ

IT 11.8 ms IT 23.3 ms

High IQ 60.14±6.48 75.26±9.09Low IQ 54.27±5.07 64.21±11.13

lost from the EEG analysis, leaving 21 high IQ participants and19 low IQ participants. Therewere two factors to the behavioraldata in this experiment; one between-groups, GROUP (high vslow IQ), and one within-groups, duration of stimulus presen-tation, ITT DURATION. The dependent variable was theproportion of correct discriminations at each level of stimulusduration. Greenhouse–Geisser correction was applied for allanalyses of the factor IT DURATION. Given the descriptivelyunequal gender composition in the “high” and “low IQ” groups,the effect of gender on the amplitude of the ERP componentsand proportion correct was examined in the high IQ groupusing a 2× (GENDER) 5× (IT DURATION). This analysis revealedthat there were no significant differences between males andfemales in both N1 amplitudes and proportions correct.

3. Results

3.1. Behavioral results

The reliability of the IT test was assessed by computingPearson's r between the first and second IT test. A correlation ofr38=.91, was found for both IQ groups combined for all ITTdurations, as well as for each IQ group separately for all ITTdurations (high alone: r19=.86; low alone: r17=.89). Addi-tionally the raw Raven's scores from both groups combined,correlated with the first ITT r38=.44 and with the second ITTr38=.42. This indicates that the IT task used in this experimentwas a reliable measure of mental speed.

Therewas a significantmain effect of IQ on ITT performance(GROUP: F1,38=13.29, p=.001, ηp

2=.259), with the “high IQ”group attaining a higher proportion of correct responses thanthe “low IQ” group (see Fig. 2 and Table 1). Furthermore, thesignificantmain effect of ITDURATION(F4,152=209.54,pb .001,ηp2=0.85, ε=.47) pointed to more correct responses with

longer IT durations. Finally, there was a significant interactionbetween IQ and stimulus duration (GROUP×IT DURATION:F4,152=4.23, p=.02, ηp

2=0.1), which showed that the “highIQ” group's level of accuracy increased more rapidly across thelonger IT durations compared to that of the “low IQ” group. As

groups.

IT 35.3 ms IT 47.06 ms IT 58.83 ms

84.67±9.56 90.53±7.93 94.27±6.3371.87±14.88 75.76±16.38 81.00±17.27

Page 5: Early electro-cortical correlates of inspection time task performance

Fig. 3. Grand average ERPs at left- and right-parietal sites in the high- and low-IQ groups. This shows the significant N1 amplitude increase 150 -200 ms in the highIQ group which occurs over both hemispheres. Negativity is plotted down.

374 D. Hill et al. / Intelligence 39 (2011) 370–377

the independent sample t-tests comparing each level of thefactor IT DURATION between the “high IQ” and “low IQ” groupsrevealed, the “high IQ” group scored significantly greater ateach level of stimulus duration than the “low IQ” group, thisgroup difference being only slightly more pronounced for23.3 ms and 47.06 ms IT durations (IT=11.80: t38=−3.19,p=.003, d=1.03; IT=23.30: t38=−3.45, d=1.09, p=.001;IT=35.30: t38=−3.20, p=.003, d=1.03; IT 47.06: t38=−3.5, p=.001, d=1.07; IT 58.83, t38=−3.16, p=.004.d=1.05; see Fig. 2). The proportion of correct responses forthe IT blocks at the beginning and end of the session for each IQgroup was for all IT durations similar and non-significant (allpsN .05).

Fig. 4. N1 topographies in the high- and low-IQ groups. Illustrates the topography ofleft hemisphere occurring between 150 and 200 ms after the stimulus in the high I

3.2. EEG/ERP results

For the P1 and the P2 amplitude and latency, none of themain or interaction effects turned out significant. While therewere no significant effects of IT or IQ on the latency of the N1,for the N1 amplitude there were significant main effects of ITDURATION (F4,148=5.14, p=.003, ηp

2=.12, ε=.72) and ofGROUP (F1,37=8.07, p=.007, ηp

2=.18), with longer ITdurations increasing the N1 amplitude and the “high IQ”group having significantly greater N1 amplitudes than the“low IQ” group (see Fig. 3). For the two groups combined, the11.80 ms elicited a significantly smaller N1 than the 22.60 ms(t38=3.54, p=.001, d=0.56), 35.30 ms (t38=2.32, p=.03,

the N1 in the two IQ groups, which was significantly greater negativity in theQ group compared to the low IQ group.

Page 6: Early electro-cortical correlates of inspection time task performance

375D. Hill et al. / Intelligence 39 (2011) 370–377

d=0.38), 47.06 ms (t38=3.68, p=.001, d=0.59) and the58.83 ms conditions (t38=3.4, p=.001, d=0.56). All othercomparisons between the ITT durations were non-significant(psN .05).

The significant main effect of HEMISHERE (F1,37=6.13,p=.02,ηp

2=.14) pointed to greaterN1 amplitudes over the lefthemisphere. Although there was no significant effect ofHEMISHERE in either group (Fsb2), the N1 was indeed slightlymore pronounced over the left than over the right hemispherein the “low IQ” group (F1,17=4.27, p=.054, ηp

2=.20, d=.47),an effect that just fell short of the conventional level ofsignificance but corresponded to a medium effect size.Correspondingly, group differences over the right hemisphere(F1,37=9.32, p=.004, ηp

2=.20) were about twice the effectsize of group differences over the left hemisphere (F1,37=4.68,p=.04, ηp

2=.11).

3.3. Exploration of sex differences

There was no effect of sex in the “high IQ” group or bothgroups pooled on either proportion correct (high IQ group:F1,19=0.11, p=.74, ηp

2= .006; both groups pooled:F1,37=0.03, p=.86, ηp

2=.001) or the amplitude of the N1(“high IQ” group: F1,19=0.54, p=.47, ηp

2=.03; both groupspooled: F1,37=0.10, p=.75, ηp

2=.003). On the basis of thesepost hoc analyses we conclude that the confounding of IQ andsex had in fact no bearing on our results.

There was a significant negative correlation between theproportion correct and the amplitude of the N1 both for the35.30 ms (r38=−.42, p=.009) and the 47.06 ms (r38=−.39,p=.03) IT durations, indicating that as the proportion ofcorrect responses increased so did the amplitude of the N1. Forthis analysis, the two longer (47.06, 58.83 ms) and the shortest(11.80 ms) IT durations were omitted from the correlationanalyses due tomany of the “high IQ” participants obtaining noerrors in the longer ITT durations and the “low IQ” groupperforming at chance level in the shortest ITT duration.

4. Discussion

The present study was the first to look at the relationshipbetween ITT, IQ and ERPs using the same IT stimulusdurations compared between IQ groups. We were able toreplicate the well-known correlation between ITT perfor-mance and IQ and found differences between the “low IQ”and “high IQ” groups specifically in the N1 amplitude, but notin the latency of this component or the other investigated ERPcomponents. These findings as well as their potentialimplications for research on the biological basis of psycho-metric intelligence will be discussed in turn.

Firstly, the experimental design of the presented studyrevealed that compared to the “high IQ” group the “low IQ”group made more errors at each level of stimulus duration.This effect cannot be attributed to differences in age or thelevel of education between the “high IQ” and “low IQ” groups.In this experiment an inspection time task was created withfive levels of stimulus duration, this resulted in a greaterproportion of errors in the shorter presentation durationcompared to the longer duration times. A significantly greaternumber of errors were made by the “low IQ” group at eachlevel of stimulus duration. This indicates that those with a

higher level of g, indicated by the RAPM, are also more able toaccurately perceive and process briefly presented stimuli.

Secondly, the greater amplitude of the N1 in the “high IQ”group reflects a difference in the physiological response, tothe same stimulus, of the “high IQ” brain compared to the“low IQ” brain (Fig. 3). This response appears to befunctionally significant, indicated by the significant correla-tions between ITT performance and N1 amplitude. Attendingto a spatial location is known to increase the amplitudes ofboth the P1 and the N1 wave formations in response to targetstimuli (Mangun, 1995). The invariant nature of the P1 in thisexperiment indicates that the increase in accuracy of the“high IQ” group was not driven by a greater ability to attendto a spatial location or a greater level of arousal. This wouldmean that success at the ITT is not a product of increasing thequantity of information by means of directing attention at theITT, as EEG/ERP differences between IQ groups occurred laterin processing. In this study the strength of the N1 response,rather than its latency, was the factor that predicted thesuccess or lack of success between the two IQ groups.

Previouswork on the inspection time task found that those ofa higher IQ attain the same level of accuracy as lower IQindividuals when the ITT is presented for a shorter amount oftime (Grudnik & Kranzler, 2001). This has been taken to beevidence of the time taken for one inspection to occur. Due to thesimplicity of the ITT this is all that would be required. N1amplitude is known tobe greaterwhen subjects are instructed toattend to a target stimulus (Mangun & Hillyard, 1991). The N1differences found in this study indicate that differences in theability to direct attention to the Π-stimulus play a role in theobserved difference between IQ groups. The greater ability of thehigh IQ group to focus attention to theΠ-stimulus, indicated bygreater N1 amplitudes, would increase the proportion correct, asthequality of the information accumulatedwouldbehigher. Thiswould lead to the differences observed in IT experiments wherehigh IQ individuals havebeen reported toneed less time to attaina predefined level of accuracy. Indeed, the link to a generalspeediness factor (Burns & Nettelbeck, 2003), specifically speedof visualization (O'Connor & Burns, 2003), could be explained bythe differential ability of IQ groups to deploy attention, whichwould increase the quality of briefly presented stimuli.

TheN1amplitudedifference,whichoccurredbetweengroupsin this experiment, appeared 150–200 ms after stimulus presen-tation, the topography and latency is constant with previousliterature detailing the N1 discrimination effect (Mangun &Hillyard, 1991). Source localization techniques applied to the N1indicate that it represents top-down modulation of sensoryinformation entering the ventral stream fromregions of the extrastriate cortex (Hopf et al., 2002, Gomez, Clark, Luck, Fan, &Hillyard, 1994). This area shows an increase in its metabolic ratewhen attending to the features of a stimulus (Corbetta, Miezin,Dobmeyer, Shulman,&Peterson, 1991). It is thought to reflect theselected amplification of task relevant features after informationhas moved from the primary visual cortex, which in turnfacilitates task performance (Corbetta et al., 1991).

The results of this experiment are consistent with the notionthat ITT is a measure of how attention facilitates performance inanticipation for a discrimination to be performed at a cued regionof space, as opposed to a measure of speed of processing in theabsence of top down factors. This is shown in this experiment bythe lack of latency differences in the N1 component indicating

Page 7: Early electro-cortical correlates of inspection time task performance

376 D. Hill et al. / Intelligence 39 (2011) 370–377

that the same stages of processing are reached by the “high IQ”and “low IQ” groups at the same points in time.

Increases in g may be associated with a greater ability toexert top-down control on selective regions of the extra-striatecortex, resulting in superior ability to process task relevantaspects of a stimulus after their intake and passage through theprimary visual areas. The link between the ITT, a test of mentalspeed, with RAPM, a power test, can be found in the work ofCarpenter et al. (1990). The superior performance of theirBetteraven model was in part due to its superior ability tomanage larger sets of problems in working memory. This leadto a greater RAPM score for the Betteraven as it was able to usemultiple rules simultaneously as well as implement rules that,due toWMconstraints, the Fairavenmodelwas unable to do. Inthe present study, high scores on RAPMwere accompanied byan increased N1 in response to the ITT, a discrimination task,indicating that as IQ increases so does the quantity ofinformation regarding the difference betweenmultiple stimuli.In the case of the ITT the comparison between the lengths of thelegs leads to the amplification of the difference between thetwo, resulting in theN1. As the goal of the RAPM is to formulatea rule or rules that explain the differences between theelements of the matrices, an increase in the quantity ofinformation regarding the differences between the basicphysical features would facilitate its processing in WM. As agreater quantity of information would require more time todecay there is less need to rehearse information in WM whenformulating a rule. This would result in the freeing up of WMresources that can then be used in the formulation andimplementation of a rule or rules to solve the problem.

While N1 discrimination effect is known to be larger in theleft hemisphere (Hopf et al., 2002), this is the first study to findevidence to suggest that increases in IQ may reduce thesedifferences. This was shown by a trend toward statisticalsignificance betweenhemispheres in the “low IQ” group,whichwas apparent at each ITT duration. No such trend was found inthe “high IQ” group. Jung andHaier (2007) analyzed the resultsof 37 neuroimaging studies looking at the structural andfunctional differences in high IQbrains. The data showed that inboth hemispheres Brodman areas 18 and 19, which correspondto the extra-striate cortex, were involved in intelligence. Theneural generators of theN1 are alsobelieved tobe in these areas(Hopf et al., 2002; Gomez et al., 1994). In the present study thedifferences between hemispheres were largest in the “low IQ”group. This result is consistent with the idea that bothhemispheres are involved in intelligence, and that the N1 isgreater in the left hemisphere.

As the activity associated with g is bilateral in the extrastriate regions (Jung&Haier, 2007), a decrease in gwill result ina reduction in bilateral activity. However, theN1discriminationeffect is greater in the left hemisphere (Hopf et al., 2002). In thisexperiment the greatest N1 amplitude difference foundbetween hemispheres occurred in the “low IQ” group. Thismaybedue to a reduction in the bilateral activity attributable tog while the N1 discrimination effect, despite being attenuatedin this group, was still dominate over the left hemisphere. Thiswould mean that as g increases the hemisphere differencestypically seen in the N1 discrimination effect would be maskedby the increase in activity due to g.

Previous studies examining the electrophysiology of ITThave relied upon the use of the so-called Parameter Estimation

by Sequential Testing PEST paradigm (Caryl, 1994; Zhang et al.,1989b). This involves establishing each participant's own IT asthe level at which they can achieve an 85% level of accuracy,afterwhichERPs are collected from theparticipants in responseto this stimulus presentation duration. Using this method,however, would create two confounds; stimulus duration andmental effort. The confound of stimulus duration would occurdue to the correlation between IT and IQ (Grudnik & Kranzler,2001). Subsequently those of a higher IQwould be viewing theΠ-stimulus at shorter durations than those of a lower IQ,making comparisons between the ERP's of the IQ groupsdifficult to interpret. Mental effort refers to the level at whichthe individual is taxed while undergoing a cognitive task(Doppelmayr et al., 2005). By using a level ashigh as 85% (Caryl,1994; Zhang et al., 1989b) both the low and the high IQ groupsmay not have been sufficiently taxed for individual differencesin N1 amplitude to emerge. Indeed, the results of the presentstudy, which show an increase in amplitude of the N1 in the“high IQ” group, indicate that the lack of differences found inprevious studiesmay be an artifact of using the PEST paradigm.

Thirdly, the present results are relevant also in the context ofthe neural efficiency hypothesis of Haier et al. (1988), whichwas formed following the finding that a high score on RAPM isassociated with a lower level of glucose metabolism duringcognitive tasks. Once task difficulty is controlled for bypresenting subjects with a task at which they attain apreselected level of accuracy, e.g. 85%, then these differencesinmetabolic activity are no longer present (Larson, Saccuzzo, &Brown, 1994). Additionally, studies that contradict the neuralefficiency hypothesis typically include tests of cognitive abilitythat are difficult for both high and low IQ groups (Neubauer &Fink, 2009). Indeed, due to the randomization of ITT durationsused in the current experiment, participants had to prepare toperform the ITT at the lowest duration time at each trial, whichwas taxing for both IQ groups as shown in Fig. 2.

Fourthly, a common explanation given for the trend ofhigh IQ individuals to outperform low IQ individuals on the ITtask is that they are more able to sustain attention for a longerperiod of time and so accumulate fewer errors over the sameperiod of time. This interpretation would be consistent withearlier accounts of the ITT (Mackintosh, 1986) which haveundergone revisions in themeantime (Mackintosh & Bennett,2002). While the duration of this experiment was greaterthan in most IT tasks, sustained attention is unlikely toaccount for the between group differences as there was nosignificant difference in the percentage correct between thefirst and the second IT test for either the “high IQ” or the “lowIQ” groups. If sustained attention drove the differencesbetween the groups then the low g group should haveaccumulated a greater proportion of errors in the second halfof the test.

In order to further investigate the role the N1 plays in ITand if the greater amplitude found in “high IQ” individuals isdue to top-down factors or if it is a consequence of bottom-upfactors, such as different levels of myelination between IQgroups (Posthuma, de Geus, & Boomsma, 2001), the N1amplitudes of both groups could be measured in response tothe Π-stimulus but without the instructions given for adiscrimination to be made. This would result in equalamplitudes between IQ groups if the N1 amplitude increasefound in this experiment represented differential functioning

Page 8: Early electro-cortical correlates of inspection time task performance

377D. Hill et al. / Intelligence 39 (2011) 370–377

of the high IQ brain in response to task demands. However, ifN1 amplitude was not related to performance on the ITT taskthe high IQ group should be greater than the low IQ group inthe absence of instructions to perform the discrimination.

References

American Electroencephalographic Society guidelines for standard electrodeposition nomenclature. Journal of Clinical Neurophysiology, 2. (1991).,200–202.

Bors, D. A., Stokes, T. L., Forrin, B., & Hodder, S. L. (1999). Inspection time andintelligence: Practice, strategies, and attention. Intelligence, 27, 111–129.

Burns, N. R., & Nettelbeck, T. (2003). Inspection time in the structure ofcognitive abilities: Where does IT fit? Intelligence, 31, 237–255.

Burns, N. R., Nettelbeck, T., & Cooper, C. J. (2000). Event related potentialcorrelates of some human cognitive ability constructs. Personality andIndividual Differences, 29, 157–168.

Carpenter, P. A., Just, M. A., & Shell, P. (1990). What one intelligence testmeasures: A theoretical account of the Processing in Ravens Progressivematrices test. Psychological Review, 97, 404–431.

Caryl, P. G. (1994). Early event-related potentials correlate with inspectiontime and intelligence. Intelligence, 8, 15–46.

Clark, V. P., & Hillyard, S. A. (1996). Spatial selective attention affects earlyextra-striate but not striate components of the visual evoked potential.Journal of Cognitive Neuroscience, 5, 387–402.

Colom, R., Abad, F. J., Quiroga, M. A., Shih, P. C., & Flores-Mendoza, C. (2008).Working memory and intelligence are highly related constructs, butwhy? Intelligence, 36, 584–606.

Corbetta, M., Miezin, F. M., Dobmeyer, S., Shulman, G. L., & Peterson, S. E.(1991). Selective and divided attention during visual discriminations ofshape colour and speed: Functional anatomy of positron emissiontomography. The Journal of Neuroscience, 11, 2383–2402.

Deary, I. J., & Stough, C. (1996). Intelligence and inspection time: Achievements,prospects, and problems. American Psychologist, 51, 599–608.

Doppelmayr, M., Klimesch, W., Sauseng, P., Hodlmoser, K., Stadler, W., &Hanslmayr, S. (2005). Intelligence related differences in EEG-band-power. Neuroscience Letters, 381, 309–313.

Gomez, C. M., Clark, V. P., Luck, S. J., Fan, S., & Hillyard, S. A. (1994). Sources ofattention-sensitive visual event-related potentials. Brain Topography, 7,41–51.

Grudnik, J. L., & Kranzler, J. H. (2001). Meta-analysis of the relationshipbetween intelligence and inspection time. Intelligence, 29, 523–535.

Haier, R. J., Siegel, B. V., Jr., Nuechterlein, H. H., Hazlett, E.,Wu, J. C., Heather, J. P.,et al. (1988). Cortical glucose metabolic rate correlates of abstractreasoning and attention studied with positron emission tomography.Intelligence, 12, 199–217.

Hillyard, S. A., &Munte, T. F. (1984). Selective attention to colour and locationcues. An analysis of event related potentials. Perception & Psychophysics,36, 185–198.

Hillyard, S. A., Vogel, E. K., & Luck, S. J. (1998). Sensory gain control(amplification) as a mechanism of selective attention: Electrophysio-logical and neuroimaging evidence. Philosophical Transactions of theRoyal Society of Biological Sciences, 353, 1257–1270.

Hopf, J. -M., Vogel, E., Woodman, G., Heinze, H. J., & Lucky, S. J. (2002).Localizing visual discrimination processes in time and space. Journal ofNeurophysiology, 88, 2088–2095.

Hutton, U., Wilding, J., & Hudson, R. (1997). The role of attention in therelationship between inspection time and IQ in children. Intelligence, 24,445–460.

Jung, R. E., & Haier, R. J. (2007). The parieto-frontal integration theory (P-FIT)of intelligence: Converging neuroimaging evidence. The Behavioral andBrain Sciences, 30, 135–187.

Larson, G. E., Saccuzzo, D. P., & Brown, J. (1994). Motivation: Cause orconfound in information processing/intelligence correlations? ActaPsychologica, 85, 25–37.

Luck, S. J., & Hillyard, S. A. (1994). Electrophysiological correlates of featureanalysis during visual search. Psychophysiology, 31, 291–308.

Mackintosh, N. J. (1986). The biology of intelligence? British Journal ofPsychology, 77, 1–18.

Mackintosh, N. J., & Bennett, E. S. (2002). IT, IQ and perceptual speed.Personality and Individual Differences, 32, 685–693.

Makeig, S., Westerfield, M., Jung, T. P., Covington, J., Townsend, J., Sejnowski,B., & Couchesne, E., (1999). Functionally independent components of thelate positive event-related potential during visual spatial attention.Journal of Neuroscience, 19, 2665–2680.

Mangun, G. R. (1995). Neural mechanisms of visual selective attention.Psychophysiology, 32, 4–18.

Mangun, G. R., & Hillyard, S. A. (1991). Modulations of sensory-evoked brainpotentials indicate changes in perceptual processing during visualspatial priming. Journal of Experimental Psychology. Human Perceptionand Performance, 17, 1057–1074.

Necka, E. (1992). Cognitive analysis of intelligence: The significance ofworking memory processes. Personality and Individual Differences, 13,1031–1046.

Neubauer, A. C., & Fink, A. (2009). Intelligence and neural efficiency.Neuroscience and Biobehavioral Reviews, 33, 1004–1023.

O'Connor, T. A., & Burns, N. R. (2003). Inspection time and general speed ofprocessing. Personality and Individual Differences, 35, 713–724.

Posthuma, D., de Geus, E. J. E., & Boomsma, D. I. (2001). Perceptual speed andIQ are associated through common genetic factors. Behavior Genetics, 31,593–602.

Raven, J. C., Court, J. H., & Raven, J. (1993). Manual for Raven's progressivematrices and vocabulary scales—Advanced progressive matrices. Sets I & II.Oxford: Oxford Psychologists Press Ltd.

Ritter, W., Simson, R., Vaughn, H. G., & Macht, M. (1982). Manipulation ofevent related potential manifestations of information processing stages.Science, 218, 909–911.

Schweizer, K. (1996). The speed–accuracy transition due to task complexity.Intelligence, 22, 115–128.

Schweizer, K. (2005). An overview of research into the cognitive basis ofintelligence. Journal of Individual Differences, 26, 43–51.

Sheppard, L. D., & Vernon, P. P. (2008). Intelligence and speed of informationprocessing: A review of 50 years of research. Personality and IndividualDifferences, 44, 535–551.

Vickers, D., Nettelbeck, T., & Willson, R. J. (1972). Perceptual indices ofperformance: The measurement of 'inspection-time' and 'noise' in thevisual system. Perception, 1, 263–295.

Vickers, D., Pietsch, A., & Hemingway, T. (1995). Intelligence and visual andauditory discrimination: Evidence that the relationship is not due to therate at which sensory information is sampled. Intelligence, 21, 197–224.

Vickers, D., & Smith, P. L. (1986). The rationale for the inspection time index.Personality and Individual Differences, 7, 609–624.

Vogel, E. K., & Luck, S. J. (2000). The visual N1 component as an index of adiscrimination process. Psychophysiology, 37, 190–203.

White, M. (1993). The inspection time rationale fails to demonstrate thatinspection time is a measure of speed of post sensory processing.Personality and Individual Differences, 15, 185–198.

White, M. (1996). Interpreting inspection time as a measure of the speed ofsensory processing. Personality and Individual Differences, 20, 351–363.

Zhang, Y., Caryl, P. G., & Deary, I. J. (1989a). Evoked potential correlates ofinspection time. Personality and Individual Differences, 10, 379–384.

Zhang, Y., Caryl, P. G., & Deary, I. J. (1989b). Evoked potentials, inspectiontime and intelligence. Personality and Individual Differences, 10,1079–1094.


Recommended