don’t look now: the magic of...

18
Elsevier AMS Ch33-I044980 Job code: EMAW 14-2-2007 1:13 p.m. Page:697 Trimsize:165×240 MM Basal Fonts:Times Margins:Top:4.6 pc Gutter:4.6 pc Font Size:10/12 Text Width:30 pc Depth:43 Lines Chapter 33 DON’T LOOK NOW: THE MAGIC OF MISDIRECTION BENJAMIN W. TATLER University of Dundee GUSTAV KUHN University of Durham Eye Movements: A Window on Mind and Brain Edited by R. P. G. van Gompel, M. H. Fischer, W. S. Murray and R. L. Hill Copyright © 2007 by Elsevier Ltd. All rights reserved.

Upload: truongtruc

Post on 10-Jun-2018

220 views

Category:

Documents


3 download

TRANSCRIPT

Elsevier AMS Ch33-I044980 Job code: EMAW 14-2-2007 1:13p.m. Page:697 Trimsize:165×240MM

Basal Fonts:Times Margins:Top:4.6pc Gutter:4.6pc Font Size:10/12 Text Width:30pc Depth:43 Lines

Chapter 33

DON’T LOOK NOW: THE MAGIC OF MISDIRECTION

BENJAMIN W. TATLER

University of Dundee

GUSTAV KUHN

University of Durham

Eye Movements: A Window on Mind and BrainEdited by R. P. G. van Gompel, M. H. Fischer, W. S. Murray and R. L. HillCopyright © 2007 by Elsevier Ltd. All rights reserved.

Elsevier AMS Ch33-I044980 Job code: EMAW 14-2-2007 1:13p.m. Page:698 Trimsize:165×240MM

Basal Fonts:Times Margins:Top:4.6pc Gutter:4.6pc Font Size:10/12 Text Width:30pc Depth:43 Lines

698 B. W. Tatler and G. Kuhn

Abstract

How do magicians misdirect their audiences? We recorded eye movements as observerswatched a magician perform a trick on a live one-to-one basis. All observers watched thetrick twice. Half of the observers were informed in advance that they would be watchinga trick; half were not. Observers tended to follow the magician’s gaze, particularly inthe second half of the trick. Even informed observers were susceptible to the magician’ssocial cues for joint attention, following his gaze during the trick. While knowing thatthey would be watching a trick was not sufficient for observers to defeat the magician’smisdirection, watching the trick a second time was; all observers were able to describehow the magician made a cigarette disappear after viewing the trick a second time.Our findings not only demonstrate an everyday example of inattentional blindness, butalso that social cues for joint attention provide the magician with a powerful means ofmisdirecting his audience successfully.

Elsevier AMS Ch33-I044980 Job code: EMAW 14-2-2007 1:13p.m. Page:699 Trimsize:165×240MM

Basal Fonts:Times Margins:Top:4.6pc Gutter:4.6pc Font Size:10/12 Text Width:30pc Depth:43 Lines

Ch. 33: Don’t Look Now: The Magic of Misdirection 699

Picture this: you are at a magic show and the magician announces that he is going tomake a donkey appear behind a curtain in the middle of the stage. He walks to the curtain,which is lying on the floor, and raises it above his head. A moment later he drops thecurtain to the ground to reveal a real live donkey behind it, to the amazement of you andthe rest of the audience. How can the magician have performed such an impressive trick?

Now let us watch the trick again to reveal its secret: when the magician raises thecurtain above his head, his glamorous (and probably scantily clad) assistant walks acrossthe front of the stage. As they do this, the magician replaces the curtain on the floor,walks to the side of the stage, collects a donkey and drags it over behind the curtain.The magician raises the curtain again and waits for his moment to drop it to reveal the‘magically appeared’ donkey. You are somewhat less impressed. Surely this is not thesame trick? Surprisingly, it is exactly as it was performed the first time. How could it bethat you missed such an obvious act as the magician walking over to drag a donkey onstage?

Ridiculous though it may seem, this illustrates the way in which many magic tricksare performed: the magician diverts the observers by directing their attention to a dis-tracting act, while at the same time performing what would be an otherwise obviousact (see Lamont & Wiseman, 1999). Magicians have for many years accomplished theirmisdirection by combining processes that psychologists have learnt about only recently:our tendency to look where others look – social attention – and our inability to spotrather obvious events under certain circumstances – change blindness and inattentionalblindness.

The fallibility of our visual sense in detecting unexpected events has become thefocus of particular interest in recent years. Observers can fail to notice what wouldseem otherwise to be a very large change to a complex scene provided that change isaccompanied by a brief interruption to viewing (for a review, see Rensink, 2002). Thesechanges can include changing the colour of an object, moving it to another position inthe scene or removing it completely. Our surprising inability to detect seemingly obviouschanges has become known as change blindness, and can be observed when changes aremade during an eye movement (e.g., Grimes, 1996; McConkie & Currie, 1996), a blink(e.g., O’Regan, Deubel, Clark, & Rensink, 2000), or an artificial flicker of the image(Rensink, O’Regan, & Clark, 1995, 1997, 2000).

However, it is not only abrupt changes occurring during periods where the scene isoccluded that can go unnoticed by observers. Unexpected events, lasting for severalseconds, can occur in full view of an observer and yet not be detected. Professionalpilots can fail to notice an aeroplane across the runway in a simulator and go on to landthrough it (Haines, 1991). Observers watching teams pass a basketball to each other canfail to notice a person in a gorilla suit (Simons & Chabris, 1999) or a woman carryingan umbrella (Neisser, 1967) walk through the midst of the players. Such failure to detectunexpected events that occur in full view has become known as inattentional blindness(see Mack & Rock, 1998), and is thought to arise because attention has been allocated toa particular task and subset of objects in the visual display.

Elsevier AMS Ch33-I044980 Job code: EMAW 14-2-2007 1:13p.m. Page:700 Trimsize:165×240MM

Basal Fonts:Times Margins:Top:4.6pc Gutter:4.6pc Font Size:10/12 Text Width:30pc Depth:43 Lines

700 B. W. Tatler and G. Kuhn

An important concern in studies of change blindness and inattentional blindness hasbeen to relate these phenomena to everyday vision and to consider their implicationsfor how vision operates normally (see Most, Scholl, Clifford & Simons, 2005, for asimilar position). Researchers have attempted to devise more realistic situations in whichto explore these phenomena (e.g., Levin & Simons, 1997; Simons & Levin, 1998).However, even these situations remain somewhat removed from ‘normal’ experience. Amore familiar situation in which these phenomena can be studied, and one that many ofus have experienced, is magic. The misdirection employed by magicians in many of theirtricks parallels inattentional blindness paradigms; it occurs seemingly in full view of theobserver, yet is not noticed.

For the magician to be successful, the audience must be misdirected when the crucialpart of the trick is performed. How does the magician do this? From an early age ourattention is strongly influenced by other people. This tendency to attend to locationsindicated by others is known as shared (or joint) attention (e.g., Tomasello, 1995, 1999).A particularly strong cue for shared attention is where somebody else is looking; we showa strong and somewhat automatic tendency to follow someone else’s gaze (e.g., Driveret al., 1999; Langton, Watt, & Bruce, 2000; Scaife & Bruner, 1975; Triesch, Teuscher,Deák, & Carlson, 2006). Indeed, under certain conditions people automatically imitateanother person’s gaze (Ricciardelli, Bricolo, Aglioti, & Chelazzi, 2002). Perhaps this ishow the magician misdirects us? If social cues strongly influence how we direct ourattention, then doing something as simple as looking away from the act that the magicianwishes to conceal, might be enough to make us attend to the wrong location and so missan act that otherwise would have been obvious.

In this chapter we consider the details of how a magician (the second author, GK)achieves his misdirection when performing a simple magic trick. The way in which thetrick was performed will now be described.

1. Our magic trick

We developed a trick in which the magician made a cigarette and lighter ‘disappear’.The trick was performed ‘live’ by the magician, in front of the observer, on a one-to-onebasis.

Figure 1 shows the typical progress of the trick. Facing the observer, the magicianasks if they would mind if he smokes. He looks at and reaches for the cigarette packetand removes a cigarette. While moving the cigarette towards his mouth with one hand(the ‘cigarette hand’), he turns to look at the lighter and reaches for this with his otherhand (the ‘lighter hand’). The cigarette is deliberately placed in the mouth the wrong wayaround. The magician then brings the lighter up to attempt to light the cigarette. Therethen follows the first misdirection by the magician. ‘Noticing’ that the cigarette is thewrong way around, the magician turns his head to the side, removes the cigarette andturns it around to replace it. Throughout this manoeuvre the magician keeps his gaze fixedon the cigarette and the hand manipulating it. While performing this manipulation of the

Elsevier AMS Ch33-I044980 Job code: EMAW 14-2-2007 1:13p.m. Page:701 Trimsize:165×240MM

Basal Fonts:Times Margins:Top:4.6pc Gutter:4.6pc Font Size:10/12 Text Width:30pc Depth:43 Lines

Ch. 33: Don’t Look Now: The Magic of Misdirection 701

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

9.0

0.0

Mag

icia

n’s

gaze

Look

ahea

d

“Do

you

min

d if

Ism

oke?

Pic

k up

ciga

rette

pack

Tak

e ci

gare

tte fr

om p

ack

Look

to c

igar

ette

pac

kLo

ok to

whe

re li

ghte

ris

on

tabl

e

Pla

ce c

igar

ette

inm

outh

/pi

ck u

p lig

hter

Look

to c

igar

ette

in m

outh

/lig

hter

han

d

Ligh

t cig

aret

telig

hter

Look

at

ciga

rette

han

das

turn

ciga

rette

Look

at (

empt

y)lig

hter

han

dLo

ok a

t (em

pty)

cig

aret

teha

ndLo

okah

ead

Look

at c

igar

ette

hand

and

ligh

ter

hand

as

pret

end

to li

ght u

p

Dro

p lig

hter

/turn

arou

nd c

igar

ette

Pre

tend

tolig

ht c

igar

ette

/m

ove

light

erha

nd to

sid

e

Ope

nem

pty

light

erha

nd

Dro

pci

gare

tte

Brin

g ci

gare

tteha

nd u

p to

face

Ope

nci

gare

tte h

and

to r

evea

l no

ciga

rette

Mag

icia

n’s

actio

ns

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

9.0

Figu

re1.

The

tric

kus

edin

the

pres

ent

stud

y.T

opro

w:

Imag

esill

ustr

atin

gva

riou

spo

ints

ofth

etr

ick.

Mid

dle

row

:W

here

the

mag

icia

nw

aslo

okin

g.T

hest

art

ofth

em

agic

ian’

sga

zeev

ent

corr

espo

nds

tow

hen

the

mag

icia

nbe

gan

his

head

turn

tow

ards

that

targ

et.B

otto

mro

w:

Wha

tth

em

agic

ian

was

doin

g.T

henu

mbe

red

vert

ical

lines

indi

cate

time

(in

seco

nds)

from

the

star

tof

the

tric

k.T

imin

gsof

the

mag

icia

n’s

actio

nsan

dga

zear

eth

eav

erag

eof

all

tria

lsre

cord

edin

this

stud

y.

Elsevier AMS Ch33-I044980 Job code: EMAW 14-2-2007 1:13p.m. Page:702 Trimsize:165×240MM

Basal Fonts:Times Margins:Top:4.6pc Gutter:4.6pc Font Size:10/12 Text Width:30pc Depth:43 Lines

702 B. W. Tatler and G. Kuhn

cigarette, the magician lowers his other hand, holding the lighter, towards the tabletopand drops the lighter onto his knees. The magician then brings his (now empty) hand backup to his face, while turning his head back towards the observer. He attempts to light the(now correctly positioned) cigarette and makes a show of being surprised that the lighteris no longer in his hand. This surprise is used to accomplish the second misdirection ofthis trick. Upon ‘noticing’ that the lighter is no longer in his hand, the magician turnshis head to the side, raises his (now empty) lighter hand and opens it while looking atit. At the same time, the magician lowers his other hand, which is holding the cigarette,towards the tabletop. When it is near the tabletop, the magician drops the cigarette ontohis knees. When dropped, the cigarette is usually about 10–15 cm above the tabletop andis therefore dropped in full view of the observer, visible for about 140 ms of its drop(calculated from the video records of the trick – see below). Shortly after he opens hisempty lighter hand, the magician brings the now empty cigarette hand back up towardhis face and turns to look at it, feigning surprise that the cigarette has also ‘disappeared’.Finally, the magician turns back to face the observer (a video clip of this trick can beviewed online at http://www.dur.ac.uk/gustav.kuhn/.)

2. The present study

In a previous report, we considered whether the success of the magician’s misdirection atthe time of the cigarette drop in the above trick was dependent upon an overt misdirectionof the eyes or a covert misdirection of attention (Kuhn & Tatler, 2005). In general,observers failed to spot how the trick was performed when they watched it for the firsttime, but all spotted how it was done when they watched the magician perform thetrick for a second time. Detecting the cigarette drop was not dependent upon where theobserver was fixating as it dropped; on trials where the drop was spotted, observers werestill often fixating the lighter hand (as would be predicted for a successful misdirection).This result implied that it was the deployment of covert visual attention rather than overtgaze that determined whether the observer detected the cigarette drop.

One question raised by the Kuhn and Tatler study is how the magician achieveshis misdirection. By considering the relationship between the magician’s gaze and theobserver’s gaze we can evaluate the role of social cues (shared or joint attention) duringthe trick and whether these form the basis for the magician’s success at misdirection. Eachparticipant watched the magician perform the trick twice. This allowed us to considerwhether different gaze strategies are employed to observe the trick for the second time.Performing the trick a second time also allows us to consider whether the observer canovercome the normal tendency for joint attention and not attend to the locations intendedby the magician. We divided our participants into two groups: half were told beforethe experiment that they were about to watch a magic trick in which a cigarette andlighter would be made to disappear and that they should try to work out how this wasdone. The remaining observers did not know that they were about to watch a magictrick. Given that task can have such a profound effect upon where observers fixate when

Elsevier AMS Ch33-I044980 Job code: EMAW 14-2-2007 1:13p.m. Page:703 Trimsize:165×240MM

Basal Fonts:Times Margins:Top:4.6pc Gutter:4.6pc Font Size:10/12 Text Width:30pc Depth:43 Lines

Ch. 33: Don’t Look Now: The Magic of Misdirection 703

viewing static scenes (e.g., Buswell, 1935; Yarbus, 1967), we expected prior knowledgeto have a significant impact upon how observers viewed the performances. We expecteduninformed observers to be more prone to the misdirection and thus follow the magician’sgaze more closely than informed observers. Our hypothesis for the combined effects ofprior knowledge and repetition is that these two factors will interact such that informedobservers are more able to ignore the magician’s social cues for joint attention and thatboth groups are less prone to this misdirection on the second performance of the trick.

3. Procedure

Twenty participants (mean age = 21�65� SD = 6�3) took part in this experiment; half ofwhom knew in advance that they were about to watch the trick, half of whom did not.Each observer saw the trick performed twice.

Eye movements of the observers were recorded while they watched the magic trick,using Land’s custom-built head-mounted eye-tracker (for details, see Land 1993; Land& Lee 1994). We defined seven possible gaze target regions: the magician’s face, thecigarette hand, the cigarette packet, the cigarette, the lighter hand, other positions onthe magician’s body, and other locations in the scene (such as items on the table thatwere not related to the trick). Consecutive fixations within a single region were summedand treated as a single gaze duration. Data from two of the observers were discarded atthis stage due to poor quality of the recorded videos. The direction of the magician’sgaze was determined from the participant’s eye-movement video record, by observing themagician’s head movements throughout the performances of the trick. This measure iscrude, but can be used to achieve a reasonably good idea of what the magician is lookingat throughout the trick.

We were interested in how both prior knowledge and repetition of the trick influencedviewing strategies. Because we did not want to confound our data with any strategicdifferences that might arise in association with detecting the cigarette drop, we excludedthe data for the two observers who spotted the cigarette drop on the first performance ofthe trick. The others all spotted it on the second performance. Eight of these remainingobservers were previously informed that they would be watching the trick and eight wereuninformed.

Given that the trick was performed live to each observer, it is important to ensurethat there is a reasonable degree of consistency in the way in which this trick wasperformed. We divided the trick up into a series of easily identifiable actions carried outby the magician (depicted in Figure 1) and recorded the time at which these happenedthroughout each performance. We found no interaction between performance number (thetwo performances of the trick) and the prior knowledge of the observers on the timings ofthe actions performed by the magician throughout the trick, F�7� 8� = 0�660� p = 0�701.Similarly, we found no interaction between performance number and the prior knowledgeof the observers on the times at which the magician moved his gaze to each of the targetsdepicted in Figure 1 throughout the trick, F�7� 8� = 0�436� p = 0�866. Thus, there were no

Elsevier AMS Ch33-I044980 Job code: EMAW 14-2-2007 1:13p.m. Page:704 Trimsize:165×240MM

Basal Fonts:Times Margins:Top:4.6pc Gutter:4.6pc Font Size:10/12 Text Width:30pc Depth:43 Lines

704 B. W. Tatler and G. Kuhn

systematic differences in the way that the trick was performed to informed or uninformedobservers or on the first and second performances of the trick to each participant. Thusany differences in the strategic deployment of gaze by the observers that arise from priorknowledge or repetition of the trick will not be artefacts of the way in which the trickwas performed.

4. A typical observer

Figure 2 shows the data for one of the uninformed observers watching the first perfor-mance of the trick. There was a reasonably close association between where the magicianwas looking and where the observer fixated. The observer in Figure 2 was successfullymisdirected: she was watching the cigarette (as was the magician) while the lighter wasdropped, and was watching the lighter hand while the cigarette was dropped. We willnow consider how observers’ gaze behaviour throughout the trick and around the time ofthese two instances of misdirection is influenced by prior knowledge that they will seethe trick, or by seeing the trick for a second time.

5. Misdirection, prior knowledge, and repetition

One measure of the inspection strategies employed by observers is to consider the totalamount of time spent gazing at each of the possible gaze targets while watching thetrick. Gaze was categorised as having been directed to one of seven possible regions:the magician’s face, the cigarette hand, the cigarette packet, the cigarette, the lighterhand, other positions on the magician’s body, and other locations in the scene (such asitems on the desk that were not related to the trick). A three-way mixed design ANOVA(gaze target region, informed/uninformed, first/second performance) showed a main effectof gaze target, F�6� 9� = 49�81� p < 0�001. Table 1 summarises the simple pairwisecomparisons between the seven possible gaze target regions. No other main effects weresignificant (all p > 0�462). Observers spent most of their time looking at the magician’sface and his two hands as might be expected given the nature of the trick.

There was a two-way interaction between gaze target region and the prior knowledgeof the participant, F�6� 9� = 5�74� p = 0�010 (Figure 3a). Simple pairwise comparisonsshowed that informed participants spent longer looking at the magician’s body (otherthan his hands and face), p < 0�001. Given that the magician’s gaze is directed primarilyat either of his hands or at the observer throughout the trial (see Figure 1), fixations ofthe magician’s body might represent attempts to avoid being ‘captured’ by the magi-cian’s misdirection. Because the magician’s body is effectively the centre of the visualscene throughout the performance, fixating here might also provide an optimal locationfor viewing the trick. However, there were no other differences between informed anduninformed observers (all p > 0�096), suggesting that any strategic differences betweenthe two groups of observers were small.

Elsevier AMS Ch33-I044980 Job code: EMAW 14-2-2007 1:13p.m. Page:705 Trimsize:165×240MM

Basal Fonts:Times Margins:Top:4.6pc Gutter:4.6pc Font Size:10/12 Text Width:30pc Depth:43 Lines

Ch. 33: Don’t Look Now: The Magic of Misdirection 705

0.0

Obs

erve

r’sga

ze

Mag

icia

n’s

gaze

Mag

icia

n’s

actio

ns

Fac

eF

ace

Look

to C

P

Look

ahe

adLo

ok to

whe

re li

ghte

r is

on ta

ble

Look

to c

igar

ette

inm

outh

/ligh

ter

hand

Look

at c

igar

ette

hand

as

turn

ciga

rette

arou

nd

Look

at (

empt

y) li

ghte

rha

ndLo

ok a

t (em

pty)

cig

aret

te h

and

Look

ahea

d

Ope

nci

gare

tte h

and

to r

evea

l no

ciga

rette

Ope

nem

pty

LHD

rop

ciga

rette

Look

toC

PLo

ok a

t CH

and

LH

as

pret

end

to li

ght u

p

Fac

e

Pap

ers

LHC

igar

ette

C PC P

CP

C HF

ace

CH

LHC

HC

HLH

Fac

eF

ace

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

9.0

10.0

0.0

Key

to a

bbre

viat

ions

:

CH

= c

igar

ette

han

dC

P =

cig

aret

te p

acke

tLH

= li

ghte

r ha

nd

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

9.0

10.0

“Do

you

min

d if

Ism

oke?

Pic

kup C

PT

ake

ciga

rette

from

pac

kP

lace

cig

aret

te in

mou

th/

pick

up

light

er

Ligh

tci

gare

ttelig

hter

Dro

p lig

hter

/turn

arou

nd c

igar

ette

Pre

tend

to li

ght

ciga

rette

/mov

e lig

hter

hand

to s

ide

of fa

ce

Brin

g ci

gare

tte h

and

upto

face

Figu

re2.

Aty

pica

lob

serv

er.T

opro

w:

Whe

reth

eob

serv

erfi

xate

d.M

iddl

ero

w:

Whe

reth

em

agic

ian

look

ed.B

otto

mro

w:

Wha

tth

em

agic

ian

was

doin

g.T

henu

mbe

red

vert

ical

lines

indi

cate

time

(in

seco

nds)

from

the

star

tof

the

tric

k.T

his

figu

rede

mon

stra

tes

the

sim

ilari

tybe

twee

nw

here

the

mag

icia

nlo

oks

and

whe

reth

eob

serv

erlo

oks.

For

exam

ple,

afte

rju

stun

der

two

seco

nds,

the

mag

icia

ntu

rned

his

head

tow

ards

the

light

eron

the

tabl

e.A

ppro

xim

atel

yon

ese

cond

late

r,as

the

mag

icia

nre

ache

dfo

rth

elig

hter

and

pick

edit

up,t

heob

serv

erdi

rect

edhe

rga

zeto

the

hand

that

the

mag

icia

nw

asus

ing

topi

ckup

the

light

eran

dbo

thth

em

agic

ian

and

obse

rver

wat

ched

this

hand

asit

was

brou

ght

tow

ards

the

face

.

Elsevier AMS Ch33-I044980 Job code: EMAW 14-2-2007 1:13p.m. Page:706 Trimsize:165×240MM

Basal Fonts:Times Margins:Top:4.6pc Gutter:4.6pc Font Size:10/12 Text Width:30pc Depth:43 Lines

706 B. W. Tatler and G. Kuhn

Table 1Results of simple pairwise comparisons for the total time spent gazing at each of the seven possible gaze targetswhen watching the trick. Boldface p-values indicate differences that were significant at the < 0�050 level. Meantotal gaze time (along with the standard deviation) is given for each target region on the left of the table

FaceCigarettehand

Cigarettepacket Cigarette

Lighterhand Body Other

FaceX̄ = 2315, – .346 <0.001 0.004 0.985 <0.001 <0.001SD = 240�1

Cigarette handX̄ = 3056, – – <0.001 <0.001 0.001 <0.001 <0.001SD = 154�7

Cigarette packX̄ = 380, – – – 0.817 0.001 >0�999 >0�999SD = 103�9

CigaretteX̄ = 854, – – – – 0.115 0.659 0.030SD = 149�1

Lighter handX̄ = 1575, – – – – – 0.001 <0.001SD = 150�1

BodyX̄ = 450, – – – – – – 0.183SD = 72�7

OtherX̄ = 157, – – – – – – –SD = 63�7

Gaze target region

(a) (b)

Tot

al g

aze

time

durin

gpe

rfor

man

ce/m

s

0

500

1000

1500

2000

2500

3000

3500Uninformed observersInformed observers

Gaze target regionFace CH CP Cigarette LH Body Other Face CH CP Cigarette LH Body Other

Tot

al g

aze

time

durin

gpe

rfor

man

ce /m

s

0

1000

2000

3000

4000

50001st Performance of trick2nd Performance of trick

p < .001

p = .004

p < .001

Figure 3. (a) Mean total gaze times in each target region (+1 SE) for the informed and uninformed observers.(b) Mean total gaze times in each target region (+1 SE) for the two performances of the trick.

Elsevier AMS Ch33-I044980 Job code: EMAW 14-2-2007 1:13p.m. Page:707 Trimsize:165×240MM

Basal Fonts:Times Margins:Top:4.6pc Gutter:4.6pc Font Size:10/12 Text Width:30pc Depth:43 Lines

Ch. 33: Don’t Look Now: The Magic of Misdirection 707

The only other significant interaction was a two-way interaction between gaze targetregion and performance number, F�6� 9� = 3�69� p = 0�039 (Figure 3b). Simple pairwisecomparisons showed that observers spent more time fixating the magician’s face duringthe first performance of the trick than during the second performance, p = 0�004, and thatthey spent more time fixating the cigarette hand during the second performance of thetrick than during the first performance, p < 0�001. No other pairwise comparisons weresignificant (all p > 0�118). Spending more time looking at the cigarette hand and lesstime looking at the magician’s face on the second trial is an entirely plausible strategyfor working out how the magician made the cigarette ‘disappear’.

Our measure of the overall time spent gazing at a region does not account for how gazesto this region are distributed throughout the performance; for example, the 2.3 s for whichobservers watch the cigarette hand (Figure 3a) may be a single gaze at a certain point inthe trick, or may be a series of short gazes to the cigarette hand distributed throughoutthe trick. We therefore examined more directly what was fixated at various points in theperformance of the trick and whether this differed according to the prior knowledge ofthe observers or whether they were viewing the trick for the first or second time.

Figure 4 shows plots of the allocation of gaze to each of the seven possible gaze targetregions throughout the course of the performance of the trick by all observers. Whileinformed and uninformed observers were largely similar in the way they watched thetrick, there were some differences. During the first performance, uninformed observerstended mainly to fixate the cigarette hand only during a period between 2.5 and 4.5 sinto the trick and again at the end of the trick. Looks to the cigarette hand by informedobservers were less clustered around these two time periods, with more looks to thecigarette hand throughout the middle portions of the trick. This result might suggesta strategic difference in the way that informed and uninformed observers watched thetrick, with informed observers showing a tendency to monitor the cigarette over a moreextended section of the trick than uninformed observers.

The more striking differences in inspection behaviour arose between the first andsecond performances of the trick. All observers seemed to adopt a strategy of largelywatching the cigarette hand through much of the second performance of the trial. Therewere also fewer looks to the magician’s face throughout the trial.

If we consider the lighter and cigarette drops, it can be seen that the magician’smisdirection appears to be effective. Shortly before the lighter was dropped, the magicianturned his head to watch the cigarette and cigarette hand as he turned the cigarette around(see also Figure 1). At this time during the first performance of the trick, informedobservers tended to be looking at the cigarette hand as expected, but uninformed observerstended to be looking at the magician’s face (Figure 4). Although the latter is not entirelyas might be expected (or as the magician might hope), both represent an effectivemisdirection from the lighter hand as the lighter is dropped.

In the case of the cigarette drop all participants tended to follow the magician’s gazeduring the first performance of the trick and be looking at the lighter hand at this time.Because both groups of observers tended to watch the cigarette hand throughout mostof the second performance of the trick, the magician’s misdirection was less effective.

Elsevier AMS Ch33-I044980 Job code: EMAW 14-2-2007 1:13p.m. Page:708 Trimsize:165×240MM

Basal Fonts:Times Margins:Top:4.6pc Gutter:4.6pc Font Size:10/12 Text Width:30pc Depth:43 Lines

708 B. W. Tatler and G. Kuhn

Face

Cigarette hand

Cigarette pack

Cigarette

Lighter hand

Body

Other

Face

Cigarette hand

Cigarette pack

Cigarette

Lighter hand

Body

Other

1st performance 2nd performance U

ninf

orm

edob

serv

ers

Info

rmed

obse

rver

s

0 2 8 10

Time/s

Lighterdrop

Cigarettedrop

Lighterdrop

Cigarettedrop

4 6

Time/s

0 2 8 104 6

Figure 4. Plots of the total number of observers looking at each of the 7 possible gaze target regions overthe course of the performance of the trick. In each plot, the baseline for each target region is 0 (no observersfixated that region at that time) and the maximum was 8 (all observers fixated that region at that time). Plotsshow these data for both uninformed (upper panels) and informed (lower panels) observers and for both thefirst (left panels) and second (right panels) performances of the trick. The timings of the lighter and cigarettedrops are also indicated by the bold dashed lines. The faint dashed lines indicate when the magician turned hishead towards the target of misdirection before the drop (i.e. to the cigarette hand just before the lighter dropand to the lighter hand just before the cigarette drop – see Figure 1).

However, it is interesting to note that when the cigarette was dropped, the observersdid again show a tendency to be misdirected to the lighter hand. Here again arises adifference between the two groups of observers. When uninformed observers watchedthe trick a second time, most fixated the lighter hand around the time of the cigarettedrop, whereas informed observers fixated either the lighter hand or the magician’s body(in equal number). Fixating the body at this time might represent a ‘disengagement’strategy; defeating the magician’s misdirection by fixating a ‘neutral’ location in thescene from which the cigarette drop might be detected. Thus although both groupsof observers successfully detected the cigarette drop on the second performance, theirviewing strategies at the time of this event appear to be somewhat different.

One interesting observation in the above results is that detecting the cigarette did notappear to be dependent upon fixating on or very near to the dropping cigarette (as wasalso found in Kuhn & Tatler, 2005); observers detected the drop in the second trial even

Elsevier AMS Ch33-I044980 Job code: EMAW 14-2-2007 1:13p.m. Page:709 Trimsize:165×240MM

Basal Fonts:Times Margins:Top:4.6pc Gutter:4.6pc Font Size:10/12 Text Width:30pc Depth:43 Lines

Ch. 33: Don’t Look Now: The Magic of Misdirection 709

when not fixating the cigarette at the time it was dropped. Our results are therefore notentirely in accordance with the traditional view that attention and eye movements arenecessarily intimately linked (e.g., Peterson, Kramer, & Irwin, 2004). Recent studies haveargued that visual encoding of a scene is largely biased by what is selected for fixation(e.g., Hollingworth & Henderson, 2002; Tatler, Gilchrist, & Land, 2005). Hollingworth,Schrock, and Henderson (2001) found a large effect of fixation position in a changedetection task, with most detections being accompanied by fixation of the changing object.From this result they suggested that fixation of the changing object was instrumental indetecting the change. Given this, our result that detection of the cigarette drop was notheavily dependent upon fixating on or near it is rather unexpected. Our data argue that,in the case of our magic trick, attention and eye position can be de-coupled, such that theeye may be misdirected but covert attention may not. This intriguing possibility clearlydeserves closer examination in the future.

6. Do observers follow the magician’s gaze?

Throughout this chapter we have alluded to the role of the magician’s gaze in misdirectingthe observer. As we mentioned in the introduction, social cues such as where a person islooking can strongly influence an observer’s direction of attention; an observer will oftenlook where someone else directs their gaze to (e.g., Ricciardelli et al., 2002). However,this is not always the case: a speaker’s gaze at their own gestures does not necessarilyresult in an eye movement to that gesture by the observer (Gullberg, 2002).

Figure 5 shows observers watching the first performance of the trick. For the firstsection of the trick, there is a relatively loose coupling between where the magician looksand where the observers look. However, from about the time that the magician turns thecigarette around a much closer association develops. This close association between themagician’s gaze and the observer’s gaze is consistent with what is known from the socialattention literature, where people show a strong tendency to follow another person’sgaze (e.g., Driver et al., 1999; Langton et al., 2000; Scaife & Bruner, 1975; Trieschet al., 2006). In current work, our group has been exploring in more detail the issue ofsocial attentional cuing in magic. By varying the availability of social cues for sharedattention we have found that the effectiveness of a magic trick varies; the fewer socialcues provided by the magician, the less effective the trick (Kuhn & Land, 2006). It isinteresting that the close association between the gazes of the magician and the observeris not maintained throughout the trick but only develops for the second half of theperformance, when the crucial acts of the trick (the disappearances) occur. What (if any)extra cues for directing the observer’s attention are employed by the magician during thishalf of the trick that reinforce the shared attention are not entirely clear and warrant furtherinvestigation.

In order to explore the congruence between the magician’s and observer’s gaze in moredetail, we divided the trick into 10 sections of equal duration and calculated the proportionof time for which the magician and observer were looking at the same gaze target. Because

Elsevier AMS Ch33-I044980 Job code: EMAW 14-2-2007 1:13p.m. Page:710 Trimsize:165×240MM

Basal Fonts:Times Margins:Top:4.6pc Gutter:4.6pc Font Size:10/12 Text Width:30pc Depth:43 Lines

710 B. W. Tatler and G. Kuhn

Magician’sgaze LHCPFace CH/LH CH/C CH/

LHLH CH

Face

Cigarette hand

Cigarette

Cigarette pack

Lighter hand

Body

Other

Face

Cigarette hand

Cigarette

Cigarette pack

Lighter hand

Body

Other

Cigarettedrop

Lighterdrop

Uninformedobservers

Informedobservers

Key to abbreviations:

CH = Cigarette hand CP = Cigarette packet C = Cigarette LH = Lighter hand

Time

Figure 5. Where observers fixated during the first performance of the trick for uninformed (upper) and informed(lower) observers (plotted using the same conventions as in Figure 4 above). The superimposed white and greybands indicate where the magician was looking throughout the trial and therefore where the observer mighthave been expected to fixate if following the magician’s gaze. The dashed lines indicate when the lighter andcigarette drops occurred.

the observer’s gaze will necessarily lag the magician’s even if they are entirely followinghis gaze, we offset the observer’s gaze by 500 ms. A three-way mixed design ANOVA(section in trick, informed/uninformed, first/second performance) showed a main effectof the section within the trial, F�9� 126� = 21�05� p < 0�001. There was no main effect ofprior knowledge, F�1� 14� = �29� p = 0�597, underlining our previous suggestion that priorknowledge has relatively little influence on visual inspection behaviour when watchingthe trick. There was a significant two-way interaction between prior knowledge and thesection of the trick, F�9� 126� = 2�41� p = 0�015. Simple pairwise comparisons showedtwo significant differences between informed and uninformed observers: congruence washigher for informed observers in the fourth section of the trick, p = 0�009 (just beforethe lighter drop), and for uninformed observers in the seventh section, p = 0�037 (justbefore the cigarette drop).

There was a significant three-way interaction between prior knowledge, first or secondperformance of the trick, and section of the trick, F�9� 126� = 3�51� p = 0�001 (Figure 6).

Elsevier AMS Ch33-I044980 Job code: EMAW 14-2-2007 1:13p.m. Page:711 Trimsize:165×240MM

Basal Fonts:Times Margins:Top:4.6pc Gutter:4.6pc Font Size:10/12 Text Width:30pc Depth:43 Lines

Ch. 33: Don’t Look Now: The Magic of Misdirection 711

Section of trick

(a) Uninformed observers

1 2 3 4 5 6 7 8 9 10

Con

grue

nce

betw

een

mag

icia

nan

d ob

serv

er

0.0

0.2

0.4

0.6

0.8

1.01st Performance2nd Performance

1st Performance2nd Performance

1 2 3 4 5 6 7 8 9 10

Con

grue

nce

betw

een

mag

icia

nan

d ob

serv

er

Section of trick

(b) Informed observers

0.0

0.2

0.4

0.6

0.8

1.0

Figure 6. (a) Mean congruence between the magician’s gaze and observer’s gaze (±1 SE) for uninformedobservers when watching the two performances of the trick for each of ten equal duration sections of thetrick. (b) Mean congruence between the magician’s and observer’s gaze (±1 SE) for informed observers whenwatching the two performances of the trick for each of ten equal duration sections of the trick.

Simple pairwise comparisons showed one difference that approached significance betweenthe two performances of the trick for informed observers: congruence was higher in theeighth section for the first performance, p = 0�053, perhaps suggesting greater disengage-ment from the magician’s misdirection at this point (around the time of the cigarette drop)during the second performance. For uninformed observers, congruence was higher forthe first performance of the trick during the third (p = 0�033; roughly when the cigaretteis initially placed in mouth), fourth (p = 0�088; just before the lighter drop) and ninth(p = 0�028; just after the cigarette drop) sections of the trick, but higher for the secondperformance of the trick during the first section (p = 0�043; the initial engagement ofthe observer by the magician). Thus, uninformed observers appeared to be less likelyto be led by the magician’s gaze during the first half of the second performance thanin the first half of the first performance. While we found differences between informedand uninformed participants and between the two performances of the trick, these wererelatively few and for most sections of the trick there was little difference between theobserver groups or performances.

The misdirection at the time of the cigarette drop is interesting because (unlike thelighter drop) the observers were generally looking at the cigarette hand until misdirectedto the lighter hand. This event can therefore allow further insights into the dynamics ofmisdirection. Figure 7 shows that, in the first trial, the magician tended to look at thelighter hand about 740 ms before dropping the cigarette from his other hand. On average,observers followed the magician’s gaze and looked at the lighter hand 400 ms after themagician. Therefore, they tended to move their eyes to the lighter hand on average 340 msbefore the cigarette was dropped. By directing his gaze to the lighter hand some threequarters of a second before dropping the cigarette, the magician appears to allow sufficienttime to ensure that the observer is likely to have shifted their gaze to the lighter handbefore the cigarette is dropped.

Elsevier AMS Ch33-I044980 Job code: EMAW 14-2-2007 1:13p.m. Page:712 Trimsize:165×240MM

Basal Fonts:Times Margins:Top:4.6pc Gutter:4.6pc Font Size:10/12 Text Width:30pc Depth:43 Lines

712 B. W. Tatler and G. Kuhn

Magician looks at lighter hand

Observer looks atlighter hand

340 ms

740 ms

Time

Magician dropscigarette

400 ms

Figure 7. The average timings between when the magician turned his gaze towards the lighter hand to misdirectthe observer, when the observer shifted their gaze to the lighter hand, and when the cigarette was dropped.

7. Conclusions

In this chapter we have seen that when watching a magician perform a magic trick live,our capacity and propensity to follow another person’s gaze appears to be an importantaspect of a magician’s success in achieving misdirection.

As we suggested in our previous article (Kuhn & Tatler, 2005), detecting the mannerof the ‘disappearances’ was not dependent upon fixating the object around the time it wasdropped. Clearly, this not only argues that detection was achieved using covert visualattention, but also highlights the possibility that overt eye position and covert attentioncan be de-coupled by observers watching this trick. Our data also raise the interestingpossibility that while the oculomotor system may still be misdirected by the magicianeven for informed observers or during the second performance, the attentional systemseems somewhat less prone to this misdirection.

There were large differences in the viewing strategies of observers on the first andsecond trials; observers appeared to be adopting a strategy of watching the cigaretteand cigarette hand throughout the second performance. Despite these differences and anapparent attempt to defeat the magician’s misdirection, observers still seemed to be proneto the magician’s misdirection, often tending to look to the now empty lighter hand whenthe cigarette was dropped.

It is surprising how little difference there was between the uninformed and informedobservers. Given such different priorities and knowledge for the two groups of observerswhen viewing the trick, more profoundly different viewing strategies might have beenexpected. Instructions given to observers before viewing a static scene can have profoundeffects upon the subsequent viewing strategies (e.g., Buswell, 1935; Yarbus, 1967), yet inthis case they appear to have little effect. Informed observers were prone to the magician’smisdirection when they watch the trick for the first time. Our finding of only smalldifferences between the strategies of uninformed and informed observers is consistent

Elsevier AMS Ch33-I044980 Job code: EMAW 14-2-2007 1:13p.m. Page:713 Trimsize:165×240MM

Basal Fonts:Times Margins:Top:4.6pc Gutter:4.6pc Font Size:10/12 Text Width:30pc Depth:43 Lines

Ch. 33: Don’t Look Now: The Magic of Misdirection 713

with our previous study that showed little difference in detection rates for spotting thecigarette drop in the same magic trick (Kuhn & Tatler, 2005). That even the informedobservers followed the magician’s gaze demonstrates the dominance of social cues forjoint attention: even when trying to defeat the magician’s misdirection these observerswere unable to defeat their propensity for joint attention.

Acknowledgements

We thank Catherine Hughes for her invaluable help in recruiting participants for theexperiment. Tom Sgouros kindly provided us with stories of unexpectedly effective stagemagic, upon which our anecdote at the start of this chapter is based. We are very gratefulto Boris Velichkovsky, Mike Land and Roger Van Gompel for their helpful commentson a previous version of this chapter.

References

Buswell, G. T. (1935). How people look at pictures: A study of the psychology of perception in art. Chicago:University of Chicago Press.

Driver, J., Davis, G., Kidd, P., Maxwell, E., Ricciardelli, P., & Baron-Cohen, S. (1999). Gaze perception triggersreflexive visuospatial orienting. Visual Cognition, 6, 509–541.

Grimes, J. (1996). On the failure to detect changes in scenes across saccades. In K. Atkins (Ed.), Perception:Vancouver studies in cognitive science (Vol. 2, pp. 89–110). New York: Oxford University Press.

Gullberg, M. (2002). Eye movements and gestures in human interaction. In J. Hyönä, R. Radach, & H. Deubel(Eds.), The mind’s eyes: Cognitive and applied aspects of eye movements (pp. 685–703). Oxford: Elsevier.

Haines, R. F. (1991). A breakdown in simultaneous information processing. In G. Obrecht & L. W. Stark (Eds.),Presbyopia research: From molecular biology to visual adaptation (pp. 171–175). New York: Plenum Press.

Hollingworth, A., & Henderson, J. M. (2002). Accurate visual memory for previously attended objects in naturalscenes. Journal of Experimental Psychology-Human Perception and Performance, 28(1), 113–136.

Hollingworth, A., Schrock, G., & Henderson, J. M. (2001). Change detection in the flicker paradigm: The roleof fixation position within the scene. Memory & Cognition, 29(2), 296–304.

Kuhn, G., & Land, M. F. (2006). There is more to magic than meets the eye. Current Biology, 16(22), R950.Kuhn, G., & Tatler, B. W. (2005). Magic and fixation: Now you don’t see it, now you do. Perception, 34(9),

1155–1161.Lamont, P., & Wiseman, R. (1999). Magic in theory. Hartfield: Hermetic Press.Land, M. F. (1993). Eye-head coordination during driving. Paper presented at the IEEE Systems, Man and

Cybernetics, Le Touquet.Land, M. F., & Lee, D. N. (1994). Where we look when we steer. Nature, 369(6483), 742–744.Langton, S. R. H., Watt, R. J., & Bruce, V. (2000). Do the eyes have it? Cues to the direction of social atention.

Trends in Cognitive Sciences, 4(2), 50–59.Levin, D. T., & Simons, D. J. (1997). Failure to detect changes to attended objects in motion pictures.

Psychonomic Bulletin & Review, 4(4), 501–506.Mack, A., & Rock, I. (1998). Inattentional blindness. Cambridge, MA: MIT Press.McConkie, G. W., & Currie, C. B. (1996). Visual stability across saccades while viewing complex pictures.

Journal of Experimental Psychology-Human Perception and Performance, 22(3), 563–581.Most, S. B., Scholl, B. J., Clifford, E. R., & Simons, D. J. (2005). What you see is what you set: Sustained

inattentional blindness and the capture of awareness. Psychological Review, 112(1), 217–242.

Elsevier AMS Ch33-I044980 Job code: EMAW 14-2-2007 1:13p.m. Page:714 Trimsize:165×240MM

Basal Fonts:Times Margins:Top:4.6pc Gutter:4.6pc Font Size:10/12 Text Width:30pc Depth:43 Lines

714 B. W. Tatler and G. Kuhn

Neisser, U. (1967). Cognitive psychology. New York: Appleton Century Crofts.O’Regan, J. K., Deubel, H., Clark, J. J., & Rensink, R. A. (2000). Picture changes during blinks: Looking

without seeing and seeing without looking. Visual Cognition, 7(1–3), 191–211.Peterson, M. S., Kramer, A. F., & Irwin, D. E. (2004). Covert shifts of attention precede involuntary eye

movements. Perception & Psychophysics, 66, 398–405.Rensink, R. A. (2002). Change detection. Annual Review of Psychology, 53, 245–277.Rensink, R. A., O’Regan, J. K., & Clark, J. J. (1995). Image flicker is as good as saccades in making large

scene changes invisible. Perception, 24(suppl.), 26–27.Rensink, R. A., O’Regan, J. K., & Clark, J. J. (1997). To see or not to see: The need for attention to perceive

changes in scenes. Psychological Science, 8(5), 368–373.Rensink, R. A., O’Regan, J. K., & Clark, J. J. (2000). On the failure to detect changes in scenes across brief

interruptions. Visual Cognition, 7(1–3), 127–145.Ricciardelli, P., Bricolo, E., Aglioti, S. M., & Chelazzi, L. (2002). My eyes want to look where your eyes are

looking: exploring the the tendency to imitate another individual’s gaze. Neuro Report, 13(17), 2259–2264.Scaife, M., & Bruner, J. S. (1975). The capacity for joint visual attention in the infant. Nature, 253, 265–266.Simons, D. J., & Chabris, C. F. (1999). Gorillas in our midst: sustained inattentional blindness for dynamic

events. Perception, 28(9), 1059–1074.Simons, D. J., & Levin, D. T. (1998). Failure to detect changes to people during a real-world interaction.

Psychonomic Bulletin & Review, 5(4), 644–649.Tatler, B. W., Gilchrist, I. D., & Land, M. F. (2005). Visual memory for objects in natural scenes: From fixations

to object files. Quarterly Journal of Experimental Psychology Section A-Human Experimental Psychology,58(5), 931–S960.

Tomasello, M. (1995). Joint attention as social cognition. In C. Moore & P. J. Dunham (Eds.), Joint attention:Its origins and role in development (pp. 103–130). Hillsdale, NJ: Lawrence Erlbaum Associates.

Tomasello, M. (1999). The cultural origins of human cognition. Cambridge, MA: Harvard University Press.Triesch, J., Teuscher, C., Deák, G., & Carlson, E. (2006). Gaze following: why (not) learn it? Developmental

Science, 9(2), 125–147.Yarbus, A. L. (1967). Eye movements and vision. New York: Plenum Press.