Perception & Psychophysics 2002, 64 (2), 266-278 Dynamic effects of the Ebbinghaus illusion in grasping: Support for a planning/control model of action SCOTT GLOVER and PETER DIXON University of Alberta, Edmonton, Alberta, Canada A distinction between planning and control can be used to explain the effects of context-induced il- lusions on actions. The present study tested the effects of the Ebbinghaus illusion on the planning and control of the grip aperture in grasping a disk. In two experiments, the illusion had an effect on grip aperture that decreased as the hand approached the target, whether or not visual feedback was avail- able. These results are taken as evidence in favor of a planning/control model, in which planning is sus- ceptible to context-induced illusions, whereas control is not. It is argued that many dissociations be- tween perception and action may better be explained as dissociations between perception and on-line control. The distinctionbetween the premovement planningof an Since Woodworth’s(1899) seminal study,much research actionand its on-linecontrolhas a long history(e.g., Jean- has gone into characterizingthese two stages of action (e.g., nerod, 1988; Keele & Posner, 1968; Woodworth, 1899). Abrams & Pratt, 1993;Elliot,Binsted,& Heath, 1999;Flash Here, we demonstrate that the earlier portions of a grasp- & Henis, 1991; Keele & Posner, 1968; Khan, Franks, & ing movement are more affected by the Ebbinghaus illu- Goodman, 1998; Meyer, Abrams, Kornblum, Wright, & sion than are the latter portions. These results provide fur- Smith, 1988; Pratt & Abrams, 1996), and some distinctions ther support for a planning/control model (Glover, 2001; between the two stages have been elucidated.For example, Glover & Dixon, 2001a,2001b, 2001d)in which planning planning appears to be a relatively slow and deliberate is more susceptible to illusions than control. The results process. The minimum time required to initiatea movement do not support a perception/action model (e.g., Aglioti, has been found to be around 250 msec (Stark, 1968). Con- DeSouza, & Goodale, 1995; Bridgeman, 1999; Bridge- versely, on-line control appears to operate relatively man, Peery, & Anand, 1997; Goodale & Milner, 1992; Mil- quickly.In contrast to the 400-msec lag between planning ner & Goodale, 1995), in which both planning and con- and control stages hypothesized by Woodworth on the trol are thought to be less susceptible to illusions than are basis of his original study, the benefits of visual and pro- perceptions. prioceptivefeedback during on-line control have more re- Woodworth (1899) was the first to demonstrate the dis- cently been found to occur in as little as 70–150 msec (e.g., tinctionbetween the premovement planningand the on-line Evarts & Vaughn, 1978; Lee & Tatton, 1975; Paulignan, control of action. Woodworth found that when participants MacKenzie, Marteniuk,& Jeannerod,1991; Smeets, Erke- were asked to draw lines of particular lengths, they re- lens, & van der Gon, 1990; Zelaznik, Hawkins, & Kissel- quired sufficient time to correct the actions on line. For ex- burgh, 1983). ample, accuracy was worse at a drawing rate of 400 msec We have hypothesized that another difference between per line than at slower rates. Furthermore, if the line- planning and control may be how each is affected by drawing task was done without vision, participants’perfor- context-induced visual illusions. In the planning/control mance at all speeds was just as poor as when the task was model (Glover, 2001; Glover & Dixon, 2001a, 2001b, done quickly with vision. Woodworth reasoned from this 2001d), context-induced illusions affect the planning that actions were composed of two stages: an “initial im- process, but not the on-linecontrolprocess. We use the term pulse” stage that reflected the premovement planning of context-inducedto refer to distortionsthat arise owing to the the action and a subsequent “current control” stage that re- visual context surrounding the target, as opposed to those flected the on-line correction of an action via feedback that result from othertypes of optical effects, such as some mechanisms. perceptualconstancymechanisms (e.g., saccadicsuppres- sion;Goodale,Pelisson,& Prablanc,1986;Prablanc & Mar- tin, 1992), exposure aftereffects (e.g., the McColloughef- Correspondence concerning this article should be addressed to fect; McCollough,1965), or the wearing of light-refracting S. Glover, Department of Psychology, Pennsylvania State University, devices (e.g., prism goggles; Redding & Wallace, 1994, 541 Moore Bldg., University Park, PA 16402 (e-mail: [email protected]). 1997). Copyright 2002 Psychonomic Society, Inc. 266 DYNAMIC ILLUSION EFFECTS IN GRASPING 267 The different effects of visual illusionson the planning will be affected by illusionsonly when the perception and and control of actions are assumed to be due to the fact that action modules interact prior to movement initiation. each stage of action uses its own visual representation. These interactions are said to be necessary when infor- The crucial difference between the two representations mation related to the target’sidentity (such as its function, pertainsto the visual context surroundingthe target. On the weight, or texture) is required in order to carry out an one hand, the visual representation used in planning in- action. corporates the visual context surrounding the target. Infor- The perception/action model makes many predictions mation about the context is crucial in planning movements similar to those of the planning/control model and is thus that avoid obstacles and/or depend on affordances involv- also supported by many of the same results (e.g., Aglioti ing the context. On the other hand, the visual representa- et al., 1995; Bridgeman et al., 1997; Daprati & Gentilucci, tion used in control operates relativelyindependentlyof the 1997; Haffenden & Goodale, 1998). However, other re- context. This allows it to operate quickly and with a high sults are more consistent with the planning/control model level of accuracy by focusing its computational power than with the perception/action model. For example, it has solely on the target itself. As a consequence of the differ- been found that illusioneffects on actionare smaller when ential role of the context in the two representations, plan- visual feedback is available to participants, as compared ning is affected by context-induced illusions, whereas with when it is not available (Gentilucci et al., 1996; control is largely immune to such effects. Glover & Dixon, 2001c; Westwood, Heath, & Roy,2000). This model provides a ready account of the reported pat- This supports the notion, inherent in the planning/control tern of illusion effects on action. For example, parameters model, that on-line control processes are responsible for of action that are the most affected by context-inducedil- the small illusion effects on actions. Conversely,any reduc- lusionsare also those that likely reflect planningprocesses, tion in illusion effects owing to the availability of visual such as reaction time (Smeets & Brenner, 1995), move- feedbackwouldseem to be inconsistentwith the perception/ menttime(Gentilucci,Chieffi, Daprati, Saetti,& Toni, 1996; action model, because both the planningand the control of Smeets & Brenner, 1995; van Donkelaar, 1999), posture actions are held to be relatively immune to illusions. choice (Glover & Dixon, 2001a), and initial hand orienta- Another line of evidencein favor of the planning/control tion (Glover & Dixon, 2001a, 2001b, 2001d). Conversely, model comes from studies examining the dynamic illusion parameters of action that are less affected by illusionsare effect (Glover & Dixon, 2001a, 2001b, 2001d). In these those that occur at or near the end of the movement and are studies, we measured the effect of an orientation illusion thus most likely to be influenced by on-line control on the orientation of the hand as participants reached out processes. These include pointing accuracy (Bridgeman, to grasp a bar. We found that the effect of the illusion on Gemmer, Forsman, & Huemer, 2000; Bridgeman, Lewis, the orientationof the hand was large early in the reach but Heit, & Nagle, 1979; Bridgeman et al., 1997), maximum decreased to near zero by the end of the reach, allowing grip aperture (Aglioti et al., 1995; Brenner & Smeets, participantsto grasp the bar without difficulty.According 1996; Daprati & Gentilucci, 1997; Haffenden & Goodale, to the planning/control model, the effect of the illusionon 1998; S. R. Jackson & Shaw, 2000; Otto-de Haart, Carey, hand orientation decreased because the illusion affected & Milne, 1999; but see Franz, Gegenfurtner, Bülthoff, & the initial planning of the movement, but not its subse- Fahle, 2000; Pavani, Boscagli, Benvenuti, Rabuffetti, & quent control. In contrast, a dynamic illusion effect would Farne, 1999), and final hand orientation(Glover & Dixon, not be predicted by a perception/action model. According 2001a, 2001b, 2001d). to this model, the illusion should have a small and consis- An alternative to the planning/control model is the tent effect throughoutthe reach, because bothplanningand perception/action model (Aglioti et al., 1995; Goodale & control make use of a context-independentaction represen- Milner, 1992; Haffenden & Goodale, 1998; Milner & tation. Goodale, 1995). This model is based on a dorsal-stream/ ventral-stream dichotomy in the visual processing path- Overview of the Present Study ways of the macaque brain (Mishkin, Ungerleider, &