untangling spatial from temporal illusions 2002-3516

Upload: usahm

Post on 04-Apr-2018

221 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/31/2019 Untangling Spatial From Temporal Illusions 2002-3516

    1/1

    TRENDSi n Neurosciences Vol.25 No.6 June 2002 293News& Comment

    http://tin s.trends.com 0166-2236/02/$ see front matt er 2002 Elsevier Science Ltd. All right s reserved. PII: S0166-2236(02)02179-3

    News& CommentNews& Comment

    Letters

    Untan gling spatial

    from tem poral

    i l lusions

    The flash -lag effect (FLE ) is a

    phenomenon in which a flash a ligned

    with a moving object appear s t o spatially

    lag behind the st imulus. In their r ecent

    review of the F LE, Krekelberg and La ppe

    conclude t ha t th e influence of differen tial

    lat encies on th e per ception of position of

    moving objects is u nden iable[1]. They

    refer to th e hypothesis tha t differences in

    physiological lat encies m ight t ra nslate

    directly into perceptua l time differen ces.

    This idea ha s enjoyed recent popularity a s

    a proposed explanation for the

    flash -lag illusion [24]. However,this is inconsistent with evidence that

    timing judgement s between flashes an d

    moving objects a re ver y accur at e [5].

    Recent r esearch su ggests tha t t he

    FLE, a s opposed to being a tem pora l

    illusion, is a consequ ence of spat ial

    inter polation [57].

    The a ppeal t o differences in

    physiological lat ency to explain th e FLE

    ru ns t he r isk of oversimplificat ion, given

    the m any shortcomings tha t hypothesis

    must contend with. First, th e different ial

    laten cy (DL) model predicts th at t heoutcome of a ra cebetween de novo

    movement a nd a flash can be changed,

    by giving the flash a hea d sta rt ; however,

    this hypothesis has been tested an d

    disproved [6]. Second, the DL model

    predicts t hat the onset t ime of a flashed

    an d m oving object will be m isperceived;

    this hypothesis has also been tested an d

    disproved [8]. Third, th e DL model runs

    into difficulties in th e flash -initiated

    para digm, in which t he flash and moving

    object appear s imult an eously: th e DL

    model predicts th at th e moving object

    will suffer th e same delay as the flash, as

    it suddenly appear s from nowhere.

    However, the cont inuous a nd flash-

    initiat ed conditions yield the sam e

    psychophysical result [6]. Fourt h,

    skepticism about th e DL explana tion of

    the F LE is war ran ted by the conspicuous

    absen ce of physiological s upport . In fact,

    as Krek elberg and Lappe acknowledge,

    th e available physiology in t he m edial

    temporal area (MT) speaks against t he

    DL model. Thus, th e differential lat ency

    hypothesis lacks a sufficient ba se of

    support , and a ny direct conn ection

    between t he timing of neur al signals and

    th e timing of perception is cast into

    doubt by a critical an alysis of the extan t

    data .

    Much of th e confusion in th e storm y

    world of flash-lag liter at ur e can be

    att ributed to a single assumption that h as

    not been critically assess ed: the

    assumption tha t a measured spatial

    differen ce can be directly tran slated in to a

    temporal difference. Having assu med

    such tra nslation, almost all reports on t he

    FLE m easu re a perceived spat ial offset

    (e.g. 1 of visual an gle), but r eport a

    put at ively corr esponding time difference

    (e.g. 70 ms).

    Our alter na tive view, if corr ect, is fat al

    to this ass um ption: even while a movingobject has a rea l-world time

    corr esponding to ea ch position, it could be

    th at th e pairin g is no longer veridical, or

    even r etained, in the r epresentat ion of the

    stimulus in t he ner vous system. Given th e

    distributed pr ocessing in th e visual

    system , a logical possibility is th at

    position inform at ion is not persisten tly

    represent ed, but instea d is only computed

    when needed. In other words, when an

    observer is ask ed wher e a m oving object

    was at a part icular m oment, a special (and

    possibly ra re) comput at ion is thenperform ed. Asm ear of spatial positions

    must be evaluat ed (deblurred) into a

    single, una mbiguous ans wer. The resu lt of

    th is compu ta tion can be non-veridical

    tha t is, a sm ear of spatial activity across

    cortex can be evaluated at some

    inter mediat e position. In this view, time

    can be stam ped with high fidelity, but th e

    position a ssociated with t hat time is th e

    result of a deblurring process tha t

    interpolates over a smear of recent

    positions [57]. The observer is only able

    to report a per ceived position after th is

    compu ta tion is complete [9]. This

    framework nat ura lly explains other

    illusions, su ch as th e Fr hlich effect, in

    which a moving object t hat appears

    suddenly is not seen in its true st art ing

    position, but instead some distance into

    the t rajectory [10]. This suggests tha t t he

    FLE is an other incar nat ion of the

    Fr hlich effect, one in which th e spat ial

    landmark ta kes on a temporal stamp as a

    resu lt of being flash ed inst ead of being

    sta tic [6].

    Spat ial int erpolation over occupied

    positions of the m oving object offers an

    explanation for the FLE t hat h as man y

    advanta ges. First, it is consistent with

    other illusions (e.g. th e Fr hlich effect).

    Second, it is consist ent with su bjects

    ability to accurat ely judge tempora l

    relationships between flashed an d

    moving objects [5]. Thir d, it does n ot

    embed the assum ption tha t a measur ed

    spatial judgement tra nslates directly

    into a temp oral illusion. Fourth , it

    accounts n at ur ally for the roun ding of

    the curve seen in Whitney and

    Mura kam is r eversal of the moving

    object, for wh ich th ey were forced to

    appeal to an additional m echanism

    (neural delay variability or a separat e

    spat iotempora l avera ging filter) [3].

    Thus, in contr ast to Krekelberg andLappes sta tement that latency

    differences un denia blyinfluence

    perception, a spat ial explanat ion could

    prove more par simonious.

    David M. Eagleman

    Terrence J. Sejnow ski

    The Salk Institu te for Biol ogical Studies,

    10010 N. Torrey Pines Road, La Jolla,

    CA 92037, USA.

    e-mail: [email protected]

    References

    1 Krekelberg, B. and Lappe, M. (2001) Neuronallatencies a nd the position of moving objects.

    Trends Neurosci. 24, 335339

    2 Purushothaman, G. et al. (1998) Moving ahea d

    thr ough different ial visual lat ency.N at ur e 396,

    424

    3 Whitney, D. and Murak ami, I. (1998) Latency

    difference, not spat ial extra polation.N at .

    N eur osci. 1, 656657

    4 Baldo, M.V. and Klein, S.A. (1995) Extra polation

    or attent ion shift?N atur e 378, 565566

    5 Eagleman , D.M. and Sejnowski, T.J. (2000)

    Latency difference versus postdiction: Response

    to Patel et al.S cience 290, 1051

    6 Eagleman , D.M. and Sejnowski, T.J. (2000)

    Motion int egration an d postdiction in visua l

    awareness. S cience287, 203620387 Eagleman , D.M. and Sejnowski, T.J. (2000) The

    position of moving objects: Respons e to

    Krekelberg et al.S cience 289, 1107

    8 Eagleman, D.M. and Sejnowski, T.J. (2001)

    The flash la g illusion is a consequence of

    adapt ive spatial filtering. S oc. N eurosci. Abstr.

    165.29

    9 Eagleman, D.M. (2001) Visual illusions

    an d n eur obiology.N at . Rev. N eurosci. 2,

    920926

    10 Frhlich, F.W. (1923) Uber die Messun g

    der Empfindungszeit [Measur ing the

    time of sensation].Z . S in nesp hys iol . 54 ,

    5878