B E H A V I O R A L A N D B R A I N S C I E N C E S ( T 9 9 4 )1 7 . 2 9 3 - 3 5 5 Printed in the United Stales o/ Ameilca

Motion perception during selfmotion: The direct versus inferential controversy revisited AlexanderH. Wertheim TNOlnstitutefot HumanFactofs,3769ZG Soestgrberg,The Nethedands El*lroilc mallt wertheim(.rizf.tno.nI

Abstract: According to the traditional inferential theory of perception, percepts of object motion or stationarity stem from an evaluation ofaffercnt rctinal signals(which encode image motion)witb thehelp ofextraretinal signals(which encode eye movements). According to direct perception theory, on thc otherhand, the percepts derive from retinally conveyed information only. Neitherview is compatible with a perceptual phenomenon that occun during visually induced sensationsof ego motion (vection). A rnodilied version ofinferential theory yields a rnodel in which the concept ofcxtraretinal signalsis replaced by that ofreference signals,which do not cncode how the eyes move in their orbits but how they movc in space.Hence reference signals are produced not only during eye movernents but also during ego motion (i.e., in response to vestibular stimulation and to retinal image flow, which may induce vection). The present theory describes the inter{acebetween self-motion and object-motion percepts. An expedmental paradigm that allows quantitative measurement ofthe magnitude and gain ofreference signalsand the size ofthejust noticeable difference (JND) between retinal and reference signals reveals that the distinction between direct and inferential theories largely depends on: (1) a mistaken beliefthat perceptual veddicality is evidence that extrarctinal information is not involved, and (2) a failure to distinguish between (the perception o0 absolute object motion in spaceand relative rnotion ofobjects with respect to each other. The model currects these errors, and provides a new, uniffed {iamework for interpreting many phenomena in the ffeld of motion perception. Kelrvordsr direct pcrccption; effercnce c 1). In terms clf Erluation 7, this rneans that to reach the PSS such a pattern should be rnoved against the eyes. If the eff'ect is strong enough, an inoerter/ Filehne illusion should be observed (the stirnulus would, when stationary, seem tcr move uith tbe eyes). No visual capture or stationaritytendency hypothesis can be ccxnpatible with such a re-

, 15 9 o

stimulus perceived as moving against the eyes

g E ft o € E l :t os s '6 o

t

motion perceived

3 g 5

'E

t -tt"li'"

t

0

c)

Motion perception

stimulus perceived as moving with the eyes

o l

to El ; 15

,

15

c) Q)

6 q, q) E

.:o

. 8 t1 0 ? l

g o

o o f

.F o

stimulus perceived as moving against the eyes

5 0

q

5

!l il to t

15

stimulus perceived as moving with the eyes

no stimulus motron perceived

0 5 1 0 1 5 2 0 2 5 3 0 3 5 4 0 4 5 5 0 e y e v e l o c i t y( d e g i s ) Figure 4. Stirnulus velocity at the two opposite thresholds (with and against the eyes)for perceiving rnotion in spaceof a large stirnulus pattern, during a pursuit eye Inoverrlent to a fixartionpoint sweeping acrossthe patterr-r,as a function of eye velocity. Ilcad is stationary in space. Uppcr panel: stirnulus continuously visible. f,ower panel, stirnulusvisible for only 300 rnsec.(Note the occurrenceof the Filehne illusion.)

sult. Various stirnulus patterns were trsed, Each consisted of a large sinusoidal grating of a particular spatial frequency. Low spatial frecluency patterns have a stronger optokinetic potential than high spatial fiequency patterns (Berthoz & Droulez 1982; Bonnet 1982; De Graaf et al. f990). Hence the fonner should create a larger visual component in the reference signal than the latter; and with very low spatial frequencies the reference signal might becorne oversized. This indeed did happen: when the patterns were made visible long enough (l sec) to generate a visual component in the reference signal, the lowest spatial frequency pattern created an inverted Filehne illusion and increasing spatial frequency reduced G...s.At the highest spatial frequency G,...s€V€il became less than I again.g It is interesting to note that when the gratings were presented only briefly (300 msec) during the pursuit eye movement, the nonnal Filehne illusion was again always observed (G..1being approximately 0.8) and spatial {requency had no effect. This was in line with expectations because such briefly visible stirnuli, whatever their spatial frequency characteristics, have no optokinetic potential. The conclusion thtrt reference signal gain can actually be rnodulated invalidates the ernpirical basis on which the cornpromise of dual mode theory rests. The point is that the empirical criterion, which makes it possible to identify whether a percept is direct or inferential, depends on B E H A V I O R AALN D B R A I NS C I E N C E S1 g g 4 \ 1 7 . 2

299

Wertheirn: Motion PercePtion the issue of perceptual veridicality, an issue closely tied to the idea that extiaretinal signals are always u'dersized' The traclitional claim of direct perception theory is that perceptual cleviations frorn reality indicate a lack of in{irrLation in the optic array, t[at is, particular invariants are incotnlilete, or have changed structurally' Such tfie "br".tt, instances clu ,-rot reflect (deficient) characteristics of "irnpoverished" picking-up rnechanistn but ;";;;;*-l the en,rirontnent, often believed to uir.,oilnfurmationln cclnditions. Ngrtnal, ecolclgilaboratory of trrtifact b" (For "r. cally relevant percepts are tfi.ught t. be veridical. in veridicality of role ,uni" discussions oi the central l9B7; Lornbardo 1972; Gyr see direct perception theory, Ulhnan f980.) In inferential theory, the extent to which percepts gain gf deviate {i6m reality reflects the extent to which the and Mack the Since l. frorn deviates extraretinal signals (see above) illusion Filehne the on (lgTiistuclies Herrnan gai' it has been assumed that extraretinal sig'als have a always has theory less than l. consequently, inferential founcl it clifficult to explain instances of really veridical perception (see, e.g., Matin l9B2)' Tfiese contradictory views have (implicitlv) led to the not clecision rule of dual rnode theory: if a percept is mediated been have n-rust it that shclws this veridical, i'ferentially, that is, with the help of (insufficient) extraretinal infoirnation; and if the percept is veridical, it tnlst have been mediated directly (for some exarnples of this reasoning, see Bridgelnan & Graziano 1989; Mack l97B; Matin f 9'82; Stark & bridgernan l9B3). The evidence from the thresh