Available online at www.sciencedirect.com

www.elsevier.com/locate/brainres

b r a i n r e s e a r c h 1 5 9 7 ( 2 0 1 5 ) 1 0 8 – 1 1 8

http://dx.doi.org/10.0006-8993/& 2014 El

nCorrespondenceTenerife, Spain.

E-mail address:

Research Report

Neurophysiological traces of the reader'sgeographical perspective associated with the deicticverbs of motion to go and to come

Manuel de Vegan, David Beltran, Enrique Garcıa-Marco, Hipolito Marrero

University of La Laguna, Spain

a r t i c l e i n f o

Article history:

Accepted 4 December 2014

This ERP study explores how participants activate their own geographical perspective,

while reading sentences describing a motion (to come or to go), or a static spatial relation (to

Available online 12 December 2014

Keywords:

Geographical perspective

Deictic verb

ERP

Sources estimation

Medial temporal cortex

Fronto-polar cortex

1016/j.brainres.2014.12.01sevier B.V. All rights rese

to: Departamento de Ps

mdevega@ull.es (M. de V

a b s t r a c t

be) referred either to the participant's current location or a distant place. The ERPs recorded

at the place names revealed that, compared to “distant places”, “close places” enhanced

ERP's components, associated with motivational relevance, in the context of the deictic

verbs of motion to come and to go, but not in the context of the static verb to be. Also, in the

context of the verbs of motion source estimation showed that “close places” elicited more

activity than “distant places” in the medial temporal cortex (around the parahippocampal

gyrus), suggesting projection of the reader's self-relevant information, or retrieval of

geographical episodic memories. Finally, sentences describing motions congruent with

the self-perspective (e.g. “going to distant place”) elicited less activation than sentences

incongruent with the self-perspective (e.g. “coming to distant place”) in the right fronto-

polar cortex and in the posterior cingulate cortex, regions generally associated with the

other's perspective or with self/other perspective conflict. These findings provide informa-

tion on the brain processes underlying readers' perspective taking, guided by the deictic

verbs of motion.

& 2014 Elsevier B.V. All rights reserved.

2rved.

icología Cognitiva, Universidad de La Laguna, Campus de Guajara, s/n, 28205 La Laguna,

ega).

1. Introduction

We continuously monitor and update where we are, keepingunder attentional focus the current location like this room,this building, this town, or this country (e.g., Damasio, 2010;Tamir and Mitchell, 2011). Moreover, in face-to-face commu-nication speakers usually take their own location as the deictic

center (Bühler, 1965), which serves as the basis for spatial

perspective taking in language. This study explores the brainresponse to the readers' geographical perspective in thecourse of ordinary comprehension of sentences involvingdeictic verbs of motion.

Spatial perspective is marked in language by meansof deictic words. Thus, English and other languages, like

b r a i n r e s e a r c h 1 5 9 7 ( 2 0 1 5 ) 1 0 8 – 1 1 8 109

Spanish, have deictic verbs like to go and to come to expressmotion from the speaker's perspective or deictic center.Consequently, to go refers to a motion away from the speak-er's location, and to come refers to a motion towards thespeaker's location. However, in narratives, typically writtenin the third person, the characters and events are explicitlyset up in a narrative scenario that differs from the reader'sdeictic center (e.g., Black et al., 1979). In this way, readers areinduced to make a spatial “deictic shift” towards the narrativeenvironment, putting themselves “in the protagonist's shoes”and disregarding their own deictic center (Segal, 1995; Zubinand Hewitt, 1995). The relevance of spatial deictic shifts forthe comprehension of narratives is clear. Taking the prota-gonist's perspective allows readers to understand deicticterms such as you, I, here, there, or the verbs to go or to comein the framework of the narrative world, allowing the track-ing of the protagonist's surroundings and emotional states,thus contributing to the reader's immersion experience(Zwaan, 2004). For instance, objects closer to the protagonistare more accessible for the reader (Glenberg et al., 1987;Bower and Morrow, 1990; De Vega, 1995). The present study,however, tries to avoid deictic shift processes in the narra-tives, allowing the reader's own geographical locus to emergeas default deictic center.

Several neuroimaging studies have reported the neuralsubstrates of perspective taking with a variety of non-verbal(Ruby and Decety, 2001; Vogeley et al., 2004; David et al., 2006;Mitchell et al., 2005) and verbal stimuli (Ruby and Decety,2004; Mano et al., 2009; D'Argembeau et al., 2007). Thesestudies reported some brain regions specifically associatedwith the self-perspective (e.g., medial prefrontal cortex,medial temporal lobe), and other regions associated withthe other-perspective (fronto-polar cortex, temporo-parietaljunction, temporal pole, superior temporal sulcus, and pos-terior cingulate cortex/precuneus), although the self- and theother-perspective considerably overlap in the brain (seereview by Legrand and Ruby, 2009). Also, EEG-based studieshave found modulations in the P300 as an index of motiva-tional relevance associated with self-reference (Zhou et al.,2010; Shi et al., 2011) and, sometimes, modulations in earliercomponents related to attentional processes (Fields andKuperberg, 2012). EEG-based source estimation usually showsactivations in the medial prefrontal and in the medial poster-ior cortex, which are comparable with fMRI results (seereview by Knyazev, 2013). Most of the studies on perspectivetaking asked participants to explicitly judge stimuli (e.g., as

Table 1 – Example of the experimental materials resulting from cplace (close, distant). All six versions shared the introductory sen

Introduction Yesterday you met a friend from your childhood. (Ayer te

Come to local She told you that she has come to Tenerife to pass her honeGo to distant She told you that she has gone to Barcelona to pass her honGo to local She told you that she has gone to Tenerife to pass her honCome todistant

She told you she has come to Barcelona to pass her honeym

Be in local She told you that she has been in Tenerife passing her honmiel.)

Be in distant She told you that she has been in Barcelona passing her hluna de miel.)

pleasant or unpleasant), simulate actions, or describe visualscenarios either from their own or from another person'sperspective. But people could also perform implicitperspective-taking operations mediated by language. Forinstance, some ERP studies have found P300 enhancementwhen participants processed incidentally self-relevant pos-sessive pronouns rather than self-non-relevant pronouns(Zhou et al., 2010; Shi et al., 2011), and other studies foundearly N1, P1 and N2 modulations for sentences referring tothe self, by means of the pronoun “you” (Fields andKuperberg, 2012).

This study aims to test for the first time the reader's brainresponse to geographical perspective taking promoted bydeictic verbs referred to a proximal motion (to come) or adistal motion (to go). We used high-density event-relatedpotentials (ERP), a method that allows for the observation ofthe time course of perspective taking as well as the estima-tion of brain sources. The experiment employed narrativeswritten in the second person, because the pronoun “you” ismore likely than “he” or “I” to induce readers to use a self-perspective to represent spatial information (Brunyé et al.,2009; Ditman et al., 2010). Each passage described a situationin which “you” met a fictitious character in an undeterminedplace, and the next sentence mentioned the character asdisplacing herself either to the participant's own geographi-cal place (e.g., Tenerife) or to a distant geographical place(e.g., Barcelona), by using either the proximal verb to come(“venir”) or the distal verb to go (“ir”). Also, non-deictic locativesentences with the verb to be (the Spanish copula “estar”),referring also to the close or to the distant geographicalplaces, were employed as control (Table 1). These sentencesjust described static locative relations that do not presupposeany deictic center or perspective.

As mentioned above, early ERP components (N1, P1 andN2) as well as the later P300 could be sensitive, in some cases,to self-relevant information. Therefore we could expectenhancement in these early and later components in pas-sages with deictic verbs combined with close-place names(the reader's own location). On the other hand, the combina-tions of deictic verb and place that are incongruent with thereader's location (“coming to distant place” and “going toclose place”) could enhance late negativity components of theERPs (N400), in comparison with conditions that are congru-ent with the reader's location (“coming to close place”and “going to distant place”). We make this predictionbecause the N400 component is a general marker of semantic

rossing the verb (to come, to go, to be) and the geographicaltence. In parenthesis the original Spanish version.

encontraste con una amiga de la infancia.)

ymoon. (Te contó que ha venido a Tenerife a pasar su luna de miel.)eymoon. (Te contó que ha ido a Barcelona a pasar su luna de miel.)eymoon. (Te contó que ha ido a Tenerife a pasar su luna de miel.)oon. (Te contó que ha venido a Barcelona a pasar su luna de miel.)

eymoon. (Te contó que ha estado en Tenerife pasando su luna de

oneymoon. (Te contó que ha estado en Barcelona pasando su

b r a i n r e s e a r c h 1 5 9 7 ( 2 0 1 5 ) 1 0 8 – 1 1 8110

incongruence (e.g. Kutas and Ferdemeier, 2011). Given thatthe verb to be does not presuppose any relative motion from agiven perspective, there should not be any modulation eitherin P300 or in any other component when the close anddistant places are compared in the context of this verb.

Another purpose of the experiment was to explore thelikely brain sources of perspective effects. We applied sourcereconstruction analyses to detect brain regions associatedwith self/other perspective in narratives. Our task demandwas simply to read for comprehension, rather than anexplicit simulation or judgment from a given perspective. Inspite of that, we expected that the implicit perspective takinggoverned by the deictic verbs would produce some brainactivations overlapping those described in the self/otherperspective-taking literature.

2. Results

2.1. Behavioral data: coherent responses

Percentage of ‘coherent’ responses was analyzed by means ofa repeated-measures ANOVA with Geographical place (dis-tant vs. close) and Verb (to go vs. to come vs. to be) as within-subjects factors. Greenhouse–Geisser correction was applied.

A main effect of Verb, F (2, 46)¼6.55, po.01, emerged. Tocome was accepted as coherent less likely (M¼78.7%) than theother two verbs, which did not differ from each other (to go:M¼87.3%; to be: M¼87.7%). This effect was, nevertheless,qualified by an interaction with Geographical place F (2,46)¼5.7, po.025. For the verb to come, F (1, 23)¼6.93, po.015,the percentage of coherent responses was higher for close

Fig. 1 – ERP waves in four representative electrodes for sentenceverb (to be), and a close or distant place name. The analyzed tim(TW1: 180–210 ms and TW2: 280–340 ms) were obtained by meawere low pass filtered at 12 Hz for graphical purposes only.

places (M¼87.1%) than for distant places (M¼70.4%). Incontrast, for both to go and to be, there was no significantdifference between close (M¼85.8% and 85.9%) and distantplaces (M¼88.8% and 89.5%). Furthermore, for distant placeconditions, to come was accepted less likely than the othertwo verbs, F (2, 46)¼8.15, po.005, whereas for close placeconditions no difference among verbs emerged.

2.2. ERP amplitude

ERP segments time-locked to geographical place nouns wereanalyzed using a two-step data-driven procedure (see Section 5for details). First, repeated-measures ANOVAs, with Geographi-cal place and Verb as factors were conducted for every datapoint in time and space. Second, consecutive time pointsrevealing interactive effects in the previous step were collapsedinto time-windows, and the whole topography for each time-window was submitted to a cluster-based randomizationapproach to contrast pairs of conditions. This two-step data-driven procedure allowed us to obtain time-windows of sig-nificant interactions between factors, and to explore the topo-graphical distribution of the differences between pairs ofconditions.

In the first step, point-wise ANOVAs yielded two time-windows showing reliable interaction effects between Verband Geographical place. In these time windows there werealso main effects of place but not of verb type. The first time-window (TW1) extended approximately between 180 and210 ms, and the second time-window (TW2) between 280and 340 ms after the place name onset. As Fig. 1 illustrates ina representative set of electrodes the anterior P2 and theposterior N1 components overlap for the first time-window,

s combining a deictic verb of motion (to go, to come) or a statice windows, signaled with vertical boxes in the top left graphns of a statistical data-driven procedure. The ERP waveforms

b r a i n r e s e a r c h 1 5 9 7 ( 2 0 1 5 ) 1 0 8 – 1 1 8 111

while central N400-like and parietal P300 components seemsto coexist for the second time-window (see arrows and labelsover waveforms). Figs. 2 and 3 illustrate the scalp distribu-tions of the differences between pairs of conditions under-lying the significant interactions in these two time-windows,obtained by the non-parametric cluster-based randomizationtests.

For the first time-window with interactive effects (180–210 ms, TW1), these randomization tests revealed significantdifferences between distant and close places for the twodeictic verbs (to go, p¼ .002, and to come, p¼ .001), but not forthe static verb, to be. Distant places showed more negativeamplitudes than close places, though with different scalp

Fig. 2 – Differential distant-close ERP waves, and differential voltawindows obtained by means of a statistical data-driven procedurin the maps correspond to the electrodes included in significan

Fig. 3 – Differential ERP waves and differential voltage maps, fordistant and close places, at the 280–340 ms time window. The whisignificant topographical clusters for each pair-wise comparison

distribution for each deictic verb: restricted to posterior sitesfor the proximal verb to come, and extended to anterior andright-posterior sites for the distal verb to go (see topographicmaps in Fig. 2). No other significant cluster arose in this timeperiod.

For the second time-window (280–340 ms, TW2), morenegative amplitudes for distant relative to close places wereobserved, again, with the two deictic verbs (to go, po.001, andto come, po.001), but not with the static verb to be (see alsoFig. 2). The distribution of these significant effects differedbetween the two deictic verbs: broadly extended across thescalp for the verb to come, and more restricted to right-anterior and central sites in the case of to go. In addition,

ge maps for the three verbs: to go, to come, to be, at two timee: TW1 (180–210 ms) and TW2 (280–340 ms). The white spotst topographical clusters for each pair-wise comparison.

the contrasts to come-to go, to come-to be, and to go-to be inte spots in the maps correspond to the electrodes included in.

b r a i n r e s e a r c h 1 5 9 7 ( 2 0 1 5 ) 1 0 8 – 1 1 8112

Fig. 3 illustrates that for distant places, the proximal verb tocome showed larger negative amplitudes than both to go(p¼ .002, over posterior sites) and to be (p¼ .002), which didnot differ from each other. However, for close places nodifference arose in any comparison between verbs.

2.3. TANOVA and source localization

There was a significant interaction between Verb and Geo-graphical place only for a time-window between 278 and318 ms after the place name onset. One-way TANOVA ana-lyses on the average activity within this interval revealed:first, dissimilar scalp topography between distant and closeplaces for the two deictic verbs of motion (to come, p¼ .002,

Fig. 4 – (A) Sources estimation for the contrast distant-close placeactivation in the left medial temporal cortex (MTC) in the contexpolar cortex (FPC) in the context of the verb to go. Also, “comingcingulate cortex (PCC) compared to “coming to close place”. (B)“coming to distant place” increased activation in the right frontplace”. The bar diagrams show the peaks of activation for the s

and to go, p¼ .049), but not for the static verb (to be, p4.4);second, dissimilar topography between the verb to come andboth the verb to go (p¼ .003) and to be (p¼ .005), but only fordistant place conditions. Consequently, the 278–318 time-window was chosen to examine the likely brain sourceunderlying the differences in topography.

Paired t-contrast on LAURA inverse solution revealed threelikely intracranial generators of the topographical differencesobserved for the critical time-window (see Fig. 4). First, therewere sources located at the medial temporal lobe (MTC),around the parahippocampal gyrus, that showed strongeractivation in response to close than distant places, but onlyin the context of the two deictic verbs: to come, maximum t[23]¼�3.57, po.005 (at BA 34: x¼�10, y¼�4, z¼�19), and to

. Distant places in comparison with close places reducedt of the deictic verbs to come and to go, and in the right fronto-to distant place” increased activation in the posterior

Sources estimation for the contrast to come-to go shows thato-polar cortex (FPC) in comparison with “going to distantignificant sources in each experimental condition.

b r a i n r e s e a r c h 1 5 9 7 ( 2 0 1 5 ) 1 0 8 – 1 1 8 113

go, maximum t [23]¼�3.52, po.005 (at BA 34: x¼�10, y¼0,z¼�14). Second, sources at the right fronto-polar cortex (FPC)appeared more strongly activated by the incongruent condi-tion “going to close place” than the congruent condition“going to distant place” (maximum t [23]¼�3.09, po.005, atBA 10: x¼29, y¼66, z¼�33), as shown in Fig. 4A. In the rightFPC sources, stronger activation was also observed for theincongruent “coming to distant place” relative to the con-gruent “going to distant place” (Fig. 4B), maximum t [23]¼4.03, po.005, (at BA 10: x¼49, y¼60, z¼�13). Finally, sourcesat the posterior cingulate cortex (PPC) showed strongeractivation for the incongruent “coming to distant place” thanthe congruent “coming to close place”: maximum t [23]¼2.96,po.01, (at BA31: x¼3, y¼�33, z¼35), as shown in Fig. 4A. Thebar graphs in Fig. 4 reveal that sources in the left MTC were ingeneral more active for close than distant-place names in thecontext of deictic verbs; however, the other two sources resultfrom reduced activity in the congruent self-perspective con-ditions: “going to distant place” (right FPC), and “coming toclose place” (PCC).

3. Discussion

Many studies in the perspective-taking literature requestparticipants either to perform social reasoning tasks aboutthe self/others mental states or to engage in self/othersvisuo-spatial perspective. By contrast, this study showed forthe first time the brain response to the spatial (geographical)perspective implicitly modulated by linguistic mechanisms,rather than explicit reasoning operations or explicit visuo-spatial tasks. Several facts are remarkable in the results.

3.1. Relevance and congruence in perspective taking

The reader's geographical location had a privileged statuswhen it was accompanied by a deictic verb describing acharacter's motion. Thus, close places, in comparison withdistant places, enhanced early ERP components over poster-ior (N1) and central sites (P2) in the context of deictic verbs. Ithas been reported elsewhere that the N1 and P2 waves aresensitive to orthographic features of words, like length andfrequency (e.g., Hauk and Pulvermüller, 2004; Proverbio et al.,2004). Given that close-place names are more frequent thandistant-place names in the participants' linguistic experience(see Section 3.3), this could have enhanced the N1/P2 waves.However, this orthographical explanation is unlikely becausein the context of the static verb to be no difference betweenclose and distant places arose for these early components.Another possible explanation is that N1/P2 effects are asso-ciated with word predictability in the context of deictic verbs.Thus, reading, “she came to…” probably rises expectationsthat the incoming word is a close-place name (e.g., Tenerife)rather than a distant-place name (e.g. Barcelona), whereasreading, “she has been in…” would not induce any particularexpectation. However, word predictability effects are usuallyreported at later temporal windows like the N400 (e.g.,Dambacher et al., 2006). Finally, the N1/P2 effects could beassociated with early top-down attentional processes to self-relevant information in the context of deictic verbs (Fields

and Kuperberg, 2012). This latter interpretation is congruentwith the effects observed in the next time window.

In the later time-window (280–340 ms) apparently therewere two ERP components with different theoretical implica-tions. First, close places, in comparison with distant places,were associated with increased P300 amplitude over centro-parietal sites. The P300 is a well-known ERP componentsensitive to unexpected events, but it is also sensitive tosalient or self-relevant stimuli (Perrin et al., 2005; Zhou et al.,2010; Shi et al., 2011). In our case, all the place names wereequally expectable, because they were mentioned an equalnumber of times across the experiment, and thus the mostplausible interpretation is that the P300 we found is a brainresponse to the relevance of close places. Importantly, thisP300 modulation was only found in the context of the twodeictic verbs, to come and to go. Second, in the same 280–340 ms time-window, there was a negative-going wave withsimilar morphology and distribution than the typical N400usually associated with semantic incongruence, and in ourcase also partially sensitive to perspective incongruence.Specifically, “coming to distant place” (incongruent with thereader's perspective) elicited larger negativity-going wavesover posterior sites than “going to distant place” (congruentwith the reader's perspective), as shown in Fig. 3A. However,the contrast between “coming to close place” (congruent withthe reader's perspective) and “going to close place” (incon-gruent with the reader's perspective) did not produce sig-nificant effects. This ERP pattern was paralleled by thebehavioral data: participants judge “coming to distant place”as less coherent than any other condition; however, theyjudge equally coherent “going to close place” and “coming toclose place”. One possible explanation for these partial con-gruence effects is that readers simply are more familiar withclose-place names than distant-place names. However, thisfamiliarity factor alone does not explain our findings, becausein the context of the static verb, no ERP or behavioraldifferences emerged between distant and close places. More-over, the contrasts between distant and close places showeddifferent distributions on the scalp in the context of to comeand to go (Fig. 2), indicating that the two deictic verbs induceperspective-taking operations relying on different neuralnetworks, as will be explained later.

3.2. Source reconstruction in perspective taking

The source reconstruction analyses in a later time-windowfurther contribute to support perspective-taking interpreta-tions of our data. Close places, in comparison with distantplaces, elicited a broad cluster of activations in the left MTC,BA 34, around the parahippocampal gyrus (Fig. 4A). Thisregion has been associated, along with other posterior medialregions (precuneus and posterior cingulate cortex), with self-perspective, self-projection, and self-related episodic mem-ories (D’Argembeau et al., 2007; Spreng et al., 2008; Buckerand Carroll, 2007). Beyond self-reference, the MTC plays animportant role in the processing of spatial and episodicmemories (Burgess et al., 2002; Hassabis and Maguire, 2007).Therefore, the processing of deictic close-place passages,could be related to the retrieval of episodic memories corre-sponding to the readers' own geographical location (e.g.,

b r a i n r e s e a r c h 1 5 9 7 ( 2 0 1 5 ) 1 0 8 – 1 1 8114

Tenerife). This retrieval of geographical memories would beless conspicuous in the processing of deictic distant-placepassages (e.g., Barcelona), which in fact reduce activation inthe MTC (Fig. 4A). Again, it is important to recall that thereader's current location differentially activates the MTC insentences with deictic verbs, but not in sentences with thestatic verb, indicating that perspective-taking processes andthe retrieval of episodic and spatial memories require adeictic context in addition to the name of the reader's currentlocation.

The combinations of deictic verbs and places that arecongruent with the reader's own perspective generate deac-tivation in some specific brain regions. Thus, “going to distantplace” reduced the activation in the right FPC, compared tothe incongruent sentences: “going to close place” (Fig. 4A) and“coming to distant place” (Fig. 4B). The FPC has been asso-ciated with the simulation of another person's perspective(Bucker and Carroll, 2007; d’Argembeau et al., 2007; Raposoet al., 2011; Spreng et al., 2008). This fact could explain thecurrent results: when the linguistic cues of verb and placeconflict, readers could shift to the other person's perspective(the third-person character in the story), recruiting the FPC,whereas when sentences are congruent with the self-per-spective, the FPC would remain less active. Another possibi-lity is that the fronto-polar activity reflects the conflict ofperspectives, when participants try to inhibit their pre-potentand automatic self-perspective, to allow shifting to the other-person perspective (e.g., Samson et al., 2005; Ruby andDecety, 2003; van der Meer et al., 2011). On the other hand,the congruent “coming to close place” also reduced theactivation in the PCC (BA31), in comparison with the incon-gruent “coming to distant place” (Fig. 4A). This region hasbeen associated in some studies with the change to theother's perspective (d’Argembeau et al., 2007; Ruby andDecety, 2001, 2003; Mano et al., 2009). Our results fit well thisinterpretation: in the congruent condition readers keep theirown perspective, whereas in the incongruent one they recruitthe PCC to take the other's perspective.

Spatial knowledge about our surroundings is part of ourself (Damasio, 2010). Knowing and updating our surroundingsseems to be an automatic process (in most situations we“know” where we are without effort), probably more relatedto the pre-reflective than to the reflective self (Damasio, 2010;Esslen et al., 2008; van der Meer et al., 2010). Therefore, it isnot surprising that in our narratives, written in the secondperson and without an explicit characters' location, a robustself-location bias emerged, in the context of deictic verbs of

Table 2 – Frequency of verb-place combinations in the local new

To come (venir)

He/she came/has come to Tenerife (vino/ha venido a Tenerife)He/she came/has come to Barcelona (vino/ha venido a Barcelona)

To go (ir)He/she went/has gone to Tenerife (fue/ha ido a Tenerife)He/she went/has gone to Barcelona (fue/ha ido a Barcelona)

To be (estar)He/she was/has been in Tenerife (estuvo/ha estado en Tenerife)He/she was/has been in Barcelona (estuvo/ha estado en Barcelona)

motion. This bias is more evident in sentences combining theproximal deictic verb to come and a close geographical name.For their part, deictic sentences incongruent with the self-location (“going to close place” and “coming to distant place”)could promote a deictic shift to the narrative character'sperspective, or a conflict between self/other perspectives,demanding additional neural resources in the FPC or in thePCC. This study demonstrated that deictic verbs efficientlyguide readers' implicit perspective taking, generating neuro-physiological traces that partially converge with thosereported in other perspective-taking experiments, in spite ofusing quite different materials and task demands.

3.3. Limitations of this study

Could the current results be explained by differences inlexical frequency or other statistical features of language?The Spanish dictionary of word frequency (Alameda andCuetos, 1995) shows the following verb frequency per millionwords: to go (ir)¼253, to come (venir)¼62 and to be (estar)¼323. Except for the relatively low frequency of venir the verbsdid not differ much in frequency. Concerning the placenames used in this study, there is no doubt that close placesare more frequently mentioned in the participants' environ-ment than distant places. Thus, a Google search in the 75-years old local newspaper El Día obtained 731.000 appear-ances for the close place Tenerife, and only 64.000 for thedistant place Barcelona. However, these lexical factors alonedo not explain our results, because we obtained an interac-tion between verb and geographical place, in addition to theirmain effects. For instance, the difference between close anddistant places was obtained in the ERP analysis, in thecontext of the deictic verbs of motion, which induce perspec-tive taking, but not in the context of the static verb, whichonly describes a locative relation.

Another statistical feature of language that could haveinfluenced our results is the fact that some verb-placecombinations may occur more frequently than others in theparticipants' environment. For instance, our participantscould have been more frequently exposed to “she came toTenerife” than to “she came to Barcelona”. A simple Googlesearch in the local newspaper El Día provides the tokenfrequency for some representative sentences used in thisstudy (Table 2).

However, the patterns of verb-place co-occurrence obse-rved in Table 2 do not entirely predict the differential brainresponses obtained here. For instance, in spite of the

spaper El Día (search in Google as: site:eldia.es “expression”).

Frequency of the sentence % of the verb use

1546 99.75 .3

12 29.329 70.7

1957 80.9463 19.1

b r a i n r e s e a r c h 1 5 9 7 ( 2 0 1 5 ) 1 0 8 – 1 1 8 115

frequency imbalance observed between the close place anddistant place sentences with the static verb “to be” there wereno behavioral or ERP differences between both conditions.Even in the cases that words co-occurrence matches theexpectations (e.g., “he/she has come to Barcelona” is lessfrequent than “he/she has gone to Barcelona”), we cannotconclude from this fact that words co-occurrence “causes”the differential performance nor brain activity in these con-ditions. It seems rather obvious that words co-occurrence isjust a reflection in language of cumulative perspective-takingepisodes in everyday communication. We suggest that whathas a causal impact in our results is the participants' trend touse their own geographical perspective by default when theyread sentences with a deictic verb of motion.

We cannot generalize the current results to other deicticverbs of motion like “to bring” or “to take”, which also involvean implicit deictic center. In a previous behavioral study, itwas found that these verbs offer similar results as “to come”and “to go” in reading times and sensibility judgments (deVega, Castillo, and Junco, 2010, unpublished), but we do notprovide here any EEG-based analysis for narratives with thesedeictic verbs of motion. On the other hand, given the fact thatthe experiment was performed with participants living in asingle place (the town of La Laguna, in Tenerife, Canarias), weonly collected a partial evidence of geographical perspectivetaking. A complementary experiment could be run with asample of participants living in Barcelona (Catalonia, Penin-sula), for whom the roles of close and distant places arereversed. Another future research line could be exploringhow fast geographical perspective is updated, when peoplehave just moved to another town. Do they take immediatelythis new town as their deictic center, or do they still keeptheir living place as default deictic center?

Finally, the EEG-based source analysis we employed hereprovided a topographic estimation of the brain activity, ratherthan a complete picture of the neural networks associatedwith perspective taking in the comprehension of deicticlanguage. The fact that the sources we found (PCC, MTCand FPC) have been identified in the fMRI literature asfunctional regions in self/other perspective-taking tasks,reinforce our results. However, we did not find any evidenceof activation in the medial prefrontal cortex, which has beenrepeatedly reported in many studies as part of the self-reference neural network. This could be due to the scarceresolution provided by the EEG-based source estimationtechnique, but could also be due to the fact that in thenarrative comprehension, the self-perspective was an impli-cit process rather than an explicit or reflective one. Thus,according to recent meta-analyses, fronto-medial regions aremainly associated with reflective or evaluative processesrather than specific self-perspective processes (van der Meeret al., 2010; Legrand and Ruby, 2009). Additional studies withhigh-resolution neuroimaging techniques will be required toclarify this point.

4. Conclusions

This study has shown that readers of second-person narrativesinvolving deictic verbs of motion mobilize their own

geographical perspective by default. The evidence for thisgeographical self-perspective comes from some ERP compo-nents (N1, P2, P300), considered sometimes as signatures of self-relevance, which were enhanced by deictic sentences referringto close places. Also a cluster of activation in the medialtemporal cortex, around the parahippocampal gyrus, obtainedfor these sentences suggests self-relevance. Finally, sentencescongruent with the self-perspective elicited less activation inthe fronto-polar cortex and in the posterior cingulate cortexthan sentences incongruent with the self-perspective, indicat-ing a reduction of self/other conflict of perspectives in theformer. Unlike in other studies, the perspective effects reportedhere occurred spontaneously in the course of ordinary lang-uage comprehension, without any explicit perspective-takinginstruction.

5. Experimental procedure

5.1. Participants

Twenty-four Spanish-speaker undergraduates (18 female; ageranging from 18 to 27 years) received course credits for theirvoluntary participation. All were right-handed and reportednormal or corrected to normal vision and no neurological orneuropsychological disorder. Three additional subjects wereexcluded because of excessive drift and eye movementartifacts.

5.2. Materials and design

A total of 180 experimental passages were written in Spanish,as described before (see Table 1). A factorial design was used,with 3 Verbs (to come, to go, to be)�2 Geographical places(close, distant). The close places were La Laguna (the town),Tenerife (the island) and Canarias (the region), and the distantplaces were Barcelona (the town), Cataluña (the region), and LaPeninsula (the Spanish mainland). All the participantsreceived experimental passages with the six place names,each place appearing in 30 passages. The geographical dis-tance between the close town and the distant town is about2500 km (1550 miles), and the corresponding regions (Catalo-nia and Canarias, respectively) do not share boundaries. Sixcounterbalanced lists, each with 180 experimental passages(30 per condition) were elaborated, in such a way that eachpassage appeared in only one condition per list, but equallyoften in all conditions across lists. Sixty filler passages with aformat similar to that of the experimental passages of Table 1were added to each list. These fillers did not include geogra-phical places, or deictic and static verbs in the secondsentence; instead, they referred to a variety of actions orevents. In half of the fillers there was semantic incongruencebetween the information in the two sentences (e.g., “A fewdays ago you had coffee with a vegetarian girl, she told youthat she usually eats big steaks”).

5.3. Procedure

Participants were randomly assigned to one of the six lists.Stimulus presentation and response collection were controlled

b r a i n r e s e a r c h 1 5 9 7 ( 2 0 1 5 ) 1 0 8 – 1 1 8116

by means of Presentation 15.1 (Neurobehavioral Systems, Inc.).Each passage started with a 500-ms central fixation cross in thecenter of the screen, followed by the first sentence, whichremained on the screen until participants pressed a button toadvance. Next, after another 500-ms fixation cross, the secondsentence was automatically presented word-by-word, accordingto a variable timing schedule computed as: 192msþ32ms�number of letters. An additional 500ms were added forsentence final words, and the geographical names were alwaysexposed for 500ms. A fixed 192ms blank screen followed eachword. After the passage's final word, a 500ms blank screen wasfollowed either by the command ‘continue’ or the question ‘is itcoherent?’ Participants judged the passage coherence by pressingthe ‘yes’ or ‘no’ assigned button, and the responses werecollected for further analysis. Coherence questions appeared in30% of the passages, to ensure participants' attention to the taskand to get behavioral measures of comprehension. The totalduration of the task was approximately 50min. Fig. 5 illustratesthe timeline for an experimental passage.

5.4. EEG recording and pre-processing

EEG and EOG signals were recorded using Ag/AgCl electrodesmounted in elastic Quick-caps (Neuromedical Supplies, Com-pumedics Inc., Charlotte). The EOG signal was measured fromtwo bipolar montages: one consisted of two electrodes placedat the outer canthus of each eye, and the other included twoelectrodes placed below and above the left eye. The EEGsignal was measured from 60 electrodes arranged accordingto the standard 10–20 system, with additional electrodesplaced at cb1/cb2 and on the left and right mastoids. AllEEG electrodes were referenced on-line to an electrode atvertex and then re-referenced off-line to linked mastoids. EEGand EOG signals were amplified at 500 Hz sampling rate usingSynamp2 amplifier (Neuroscan, Compumedics Inc., Charlotte),with low and high-pass filter set at .05 and 100 Hz, respec-tively. EEG electrode impedance was kept below 5 kΩthroughout the experiment.

EEG data pre-processing was conducted using Edit 4.5 software(Neuroscan, Compumedics Inc., Charlotte). The following trans-forms were applied to each participant’s dataset. Data wereinitially down-sampled to 250 Hz and low-pass filtered at 30 Hz.EEG segments were then extracted with an interval of 100mspreceding and 700ms following the geographical place name

Fig. 5 – Timeline of experimental trials. The spaces between wolanguage frame included the complete introduction sentence: Yencontraste con una amiga de la infancia). The approximate tranTenerife to pass her honeymoon. The last frame included either the w

onset (e.g. Barcelona). On these segments, artifact rejection wasperformed in two steps. First, trials containing activity exceedinga threshold of 770 mV at vertical and horizontal EOG and EEGchannels were automatically detected and rejected. Second, non-automatically rejected artifacts were manually removed, includ-ing trials with saccades identified over the horizontal EOGchannel. For the computation of ERPs, artifact-free segmentswere finally averaged separately for each of the five experimentalconditions. A total of 8.9% of trials were excluded because ofartifacts (mainly, eye blinks and drifts). The average percentageor artifacts per condition was: “to come close”, 8%; “to comedistant”, 9.1%; “to go close”, 8.6%; “to go distant”, 10.2%; “to beclose”, 8.6%, and “to be distant”, 8.9%. Baseline correction ofaveraged data was carried out using the 100-ms period precedinggeographical place name onset.

5.5. ERP amplitude analyses

The 180 experimental trials, including those followed byquestions, were submitted to the analyses. Specifically theERP amplitudes in response to geographical place names (e.g.Barcelona) were analyzed following a two-step procedure.First, a repeated-measures ANOVA with 3 Verbs (to go, to come,to be) and 2 Geographical places (close, distant) as within-subject factors was run for each data point and scalp site(175�60). To avoid false positives (Type 1 errors), significanteffects (po.05) were considered reliable only when observedfor at least 7 consecutive time points (428 ms), and fourneighboring channels (see Schupp et al., 2004 for a similarprocedure). Greenhouse–Geisser corrections were alwaysapplied to correct for sphericity. In a second step, timeperiods of reliable interaction effects were decomposed usingthe non-parametric cluster randomization approach imple-mented in the Matlab Toolbox Fieldtrip (for details, seeOostenveld et al., 2011). This approach was chosen becauseof its adequacy to estimate the scalp distribution of signifi-cant differences between two conditions. For each pair-wisecomparison, the average activity at each scalp site for theintervals showing reliable interaction was introduced asinput to the randomization test, giving as output a numberof spatially contiguous sites (cluster, 43) in which thedifference between the two conditions reached significance(po.05). In total, nine pair-wise comparisons were planned todecompose reliable interactions with this approach: three to

rd frames correspond to blank intervals of 192 ms. The firstesterday you met a friend from your childhood (Ayer teslation of the next frames is: She told you that she has come toord “continue” or a yes/no coherence question (30% of trials).

b r a i n r e s e a r c h 1 5 9 7 ( 2 0 1 5 ) 1 0 8 – 1 1 8 117

examine the effects of Geographical place at each Verb (“tocome to distant place” vs. “to come to close place”; “to go todistant place” vs. “to go to close place”, and “to be in distantplace” vs. “to be in close place”), and six to examine theeffects of Verb at each level of Geographical place (“to go todistant place” vs. “to come to distant place”; “to go to distantplace” vs. “to be in distant place”; “to come to distant place”vs. “to be in distant place”; “to go to close place” vs. “to cometo close place”; “to go to close place” vs. “to be in close place”;and “to come to close place” vs. “to be in close place”).

5.6. Topographic and source localization analyses

A two-step procedure was also adopted to evaluate the likelyERP brain sources of the interaction between Geographicalplace and Verb. First, using the Matlab Toolbox RAGU (Koeniget al., 2011), an omnibus TANOVA with Verb and Geographi-cal place as within-subject factors was performed to assessthe presence of topographic dissimilarities. The analysesincluded all the data points between 0 and 700 ms (175). Tocontrol for false positives, significant effects (po.05) wereconsidered reliable only when observed for at least 7 con-secutive time points (428 ms). Time intervals of reliableinteractions were further decomposed by performing addi-tional one-way TANOVAs: three to examine the effect of theGeographical place at each level of the Verb, and six toexamine the effect of the Verb at each level of Geographicalplace. Note that TANOVA statistic takes into considerationthe whole distribution of potentials across the scalp, notindividual activities in each scalp site, and identifies periodsof time in which this scalp field distribution (map topogra-phy) differs between conditions (for details, Murray et al.,2008; Koenig et al., 2011). In the current study, this approachwas adopted because of the assumption that different dis-tributions of brain sources underlie dissimilar scalp topogra-phies (e.g., Murray et al., 2008). Time intervals showingreliable TANOVA results were hence selected to furtherexplore the likely intracranial sources underlying topographicdissimilarities.

To estimate the likely brain sources, the LAURA (LocalAuto-Regressive Average, Grave de Peralta Menendez et al.,2004) distributed source estimation approach was appliedusing Cartool software (http://brainmapping.unige.ch/Cartool.php). A realistic head model based on a standard MRI (MNIbrain from Montreal Neurological Institute) was used byapplying the SMAC transformation method, which trans-forms the MRI to a best-fitting sphere. Next, 5005 solutionpoints were defined in regular distances within the graymatter (6�6�6 mm) and a 3-shell spherical lead field wascalculated for this set of solution points and for the 60 scalpelectrodes. From this lead field, LAURA current densityestimates were computed for the average ERP activity in timeintervals showing significant topographical differences.Voxel-wise t-tests were run to compare the mean LAURAsource estimates of the two conditions differing in scalptopography. To estimate the Brodmann areas which best fitthe LAURA results, the location of each maximum t-value(positive or negative) was compared with the location of theBrodmann areas in Tailarach space.

Acknowledgments

This research was funded by the Spanish Ministerio deEconomía y Competitividad (Grant SEJ2011-28679), the Agen-cia Canaria de Investigación, Innovación y Sociedad de laInformación, Neurocog Project (Grant APD 09-08), the CampusAtlántico Tricontinental, and the European Regional Devel-opment Funds to Manuel de Vega.

r e f e r e n c e s

Alameda, J.R., Cuetos, F., 1995. Diccionario de frecuencias de lasunidades linguısticas del castellano. Servicio de publicacionesde la Universidad de Oviedo, Oviedo.

Black, J.B., Turner, T.J., Bower, G.H., 1979. Point of view in narrativecomprehension, memory, and production. J. Verbal Learn.Verbal Behav. 18, 187–198.

Bower, G.H., Morrow, D.G., 1990. Mental models in narrativecomprehension. Science 247, 44–48.

Brunye, T.T., et al., 2009. When you and I share perspectives:pronouns modulate perspective taking during narrativecomprehension. Psychol. Sci. 20, 27–32.

Bucker, R.L., Carroll, D.C., 2007. Self-projection and the brain.Trends Cognit. Sci. 11, 49–57.

Buhler, K., 1965. Sprachtheorie. Gustav Fisher Verlag, Stuttgart,Germany.

Burgess, N., Maguire, E.A., O’Keefe, J., 2002. The humanhippocampus and spatial and episodic memory. Neuron 35,625–641.

Damasio, A., 2010. Self Comes to Mind. Constructing theConscious Brain. William Heinemann, London.

Dambacher, M., Kliegl, R., Hofmann, M., Jacobs, A.M., 2006.Frequency and predictability effects on event-relatedpotentials during reading. Brain Res. 1084, 89–103.

D’Argembeau, A., Ruby, P., Collete, F., Degueldre, C., Balteau, E.,Luxen, A., Maquet, P., Salmon, E., 2007. Distinct regions of themedial prefrontal cortex are associated with self-referentialprocessing and perspective taking. J. Cognit. Neurosci. 19,935–944.

David, N., Bewernick, B.H., Cohen, M.X., Newn, A., Lux, S., Fink,G.R., Shan, N.J., Vogeley, K., 2006. Neural representations ofself versus other: visual-spatial perspective taking andagency in a virtual ball-tossing game. J. Cognit. Neurosci. 18,898–910.

De Vega, M., 1995. Backward updating of mental models duringcontinuous reading of narratives. J. Exp. Psychol.: Learn. Mem.Cognit. 21, 373–385.

Ditman, T., Brunye, T.T., Mahoney, C.R., Taylor, H.A., 2010. Simulatingan enactment effect: pronouns guide action simulation duringnarrative comprehension. Cognition 115, 172–178.

Esslen, M., Metzler, S., Pascual-Marqui, R., Jancke, L., 2008. Pre-reflective and reflective self-reference: a spatiotemporal EEGanalysis. NeuroImage 42, 437–449.

Fields, E.C., Kuperberg, G.R., 2012. It’s all about you: an ERP studyof emotion and self-relevance in discourse. Neuroimage 62,562–574.

Glenberg, A., Meyer, M., Lindem, K., 1987. Mental modelscontribute to foregrounding during text comprehension. J.Mem. Lang. 26, 69–83.

Grave de Peralta Menendez, R., Murray, M., Michel, C., Martuzzi,R., Gonzalez Andino, S., 2004. Electrical neuroimaging basedon biophysical constraints. NeuroImage 21, 527–539.

Hassabis, D., Maguire, E.A., 2007. Deconstructing episodicmemory with construction. Trends Cognit. Sci. 11,299–306.

b r a i n r e s e a r c h 1 5 9 7 ( 2 0 1 5 ) 1 0 8 – 1 1 8118

Hauk, O., Pulvermuller, F., 2004. Effects of word length andfrequency on the human event-related potential. Clin.Neurophysiol. 115, 1090–1103.

Knyazev, G.G., 2013. EEG correlates of self-referential processing.Front. Hum. Neurosci. http://dxdoi.org/10.3389/fnhum.2013.00264.

Koenig, T., Kottlow, M., Stein, M., Melie-Garcıa, L., 2011. Ragu: afree tool for the analysis of EEG and MEG event-related scalpfield data using global randomization statistics. Comput.Intell. Neurosci. 2011, 938925.

Kutas, M., Federmeier, K.D., 2011. Thirty years and counting:finding meaning in the N400 component of the event-relatedbrain potential (ERP). Annu. Rev. Psychol. 62, 1–27.

Legrand, D., Ruby, P., 2009. What is self-specific? Theoreticalinvestigation and critical review of neuroimaging results.Psychol. Rev. 116, 252–282.

Mano, Y., Harada, T., Sugiura, M., Saito, D.N., Sadato, N., 2009.Perspective taking as part of narrative comprehension: afunctional MRI study. Neuropsychologia 47, 213–824.

Mitchell, J.P., Banaji, M.R., Macrae, C.N., 2005. The link betweensocial cognition and self-referential thought in the medialprefrontal cortex. J. Cognit. Neurosci. 17, 1306–1315.

Murray, M., Brunet, D., Michel, C., 2008. Topographic ERPanalyses: a step-by-step tutorial review. Brain Topogr. 20,249–264.

Oostenveld, R., Fries, P., Maris, E., Schoffelen, J.-M., 2011.FieldTrip: open source software for advanced analysis of MEG,EEG, and invasive electrophysiological data. Comput. Intell.Neurosci. 2011, 156869.

Perrin, F., Maquet, P., Peigneux, P., Ruby, P., Degueldre, C., Balteau,E., et al., 2005. Neural mechanisms involved in the detectionof our first name: a combined ERPs and PET study.Neuropsychologia 43, 12–19.

Proverbio, A.M., Vecchi, L., Zani, A., 2004. From orthography tophonetics: ERP measures of grapheme-to-phonemeconversion mechanisms in reading. J. Cognit. Neurosci. 16,301–317.

Raposo, A., Vicens, L., Clithero, J.A., Dobbins, I.J., Huettel, S.A.,2011. Contributions of frontopolar cortex to judgments aboutself, others and relations. Soc. Cognit. Affect. Neurosci. 6,260–269.

Ruby, P., Decety, J., 2001. Effect of subjective perspective takingduring simulation of action: a PET investigation of agency.Nat. Neurosci. 4, 546–550.

Ruby, P., Decety, J., 2003. What you believe versus what you thinkthey believe: a neuroimaging study of conceptual perspectivetaking. Eur. J. Neurosci. 17, 2475–2480.

Ruby, P., Decety, J., 2004. How would you feel versus how do youthink she would feel? A neuroimaging study of perspectivetaking with social emotions. J. Cognit. Neurosci. 16, 988–999.

Samson, D., Apperly, I.A., Kathirgamanathan, U., Humphreys,G.W., 2005. Seeing it my way: a case of a selective deficit ininhibiting self-perspective. Brain 128, 1102–1111.

Segal, E.M., 1995. Narrative comprehension and the role of deicticshift theory. In: Duchan, J.F., et al. (Eds.). Deixis in Narrative. ACognitive Science Perspective. Hillsdale, N.J. Lawrence.

Shi, Z., Zhou, A., Liu, P., Zhang, P., Han, W., 2011. An EEG study onthe effect of self-relevant possessive pronoun: self-referentialcontent and first-person perspective. Neurosci. Lett. 494,174–179.

Schupp, H.T., Ohman, A., Junghofer, M., Weike, A.I., Stockburger,J., Hamm, A.O., 2004. The facilitated processing of threateningfaces: an ERP analysis. Emotion 4 (2), 189–200.

Spreng, R.N., Mar, R.A., Kim, A.S.N., 2008. The common neuralbasis of autobiographical memory, prospection, navigation,theory of mind, and the default mode: a quantitative meta-analysis. J. Cognit. Neurosci. 21, 489–510.

Tamir, D.I., Mitchell, J.P., 2011. The default network distinguishesconstruals of proximal versus distal events. J. Cognit.Neurosci. 23, 2945–2955.

van der Meer, L., Costafreda, S., Aleman, A., David, A.S., 2010.Self-reflection and the brain: a theoretical review and meta-analysis of neuroimaging studies with implications forschizophrenia. Neurosci. Biobehav. Rev. 34, 935–946.

van der Meer, L., Groenewold, N.A., Nolen, W.A., Pijnenborg, M.,Aleman, A., 2011. Inhibit yourself and understand the other:neural basis of distinct processes underlying theory of mind.NeuroImage 56, 2364–2374.

Vogeley, K., May, M., Ritzl, A., Falkai, P., Zilles, K., Fink, G.R., 2004.Neural correlates of first-person perspective as oneconstituent of human selfconsciousness. J. Cognit. Neurosci.16 (5), 817–827.

Zhou, A., Shi, Z., Zhang, P., Liu, P., Han, W., Wu, H., et al., 2010. AnERP study on the effect of self-relevant possessive pronoun.Neurosci. Lett. 480, 162–166.

Zubin, D.A., Hewitt, L.E., 1995. The deictic center. A theory ofdeixis in narrative. In: Duchan, J.F. (Ed.), Deixis in Narrative. Acognitive science perspective. Erlbaum Associates, Hillsdale,NJ., Lawrence, pp. 129–155.

Zwaan, R., 2004. The immersed experiencer: toward an embodiedtheory of language comprehension. In: Ross, B.H. (Ed.), ThePsychology of Learning and Motivation. Advances in Researchand Theory. Elsevier, Boston, pp. 35–62.