Does visuomotor adaptation contribute to illusion-resistant grasping?

Abstract

Do illusory distortions of perceived object size influence how wide the hand is opened during a grasping movement? Many studies on this question have reported illusion-resistant grasping, but this finding has been contradicted by other studies showing that grasping movements and perceptual judgments are equally susceptible. One largely unexplored explanation for these contradictions is that illusion effects on grasping can be reduced with repeated movements. Using a visuomotor adaptation paradigm, we investigated whether an adaptation model could predict the time course of Ponzo illusion effects on grasping. Participants performed a series of trials in which they viewed a thin wooden target, manually reported an estimate of the target's length, then reached to grasp the target. Manual size estimates (MSEs) were clearly biased by the illusion, but maximum grip apertures (MGAs) of grasping movements were consistently accurate. Illusion-resistant MGAs were observed immediately upon presentation of the illusion, so there was no decrement in susceptibility for the adaptation model to explain. To determine whether online corrections based on visual feedback could have produced illusion-resistant MGAs, we performed an exploratory post hoc analysis of movement trajectories. Early portions of the illusion effect profile evolved as if they were biased by the illusion to the same magnitude as the perceptual responses (MSEs), but this bias was attenuated prior to the MGA. Overall, this preregistered study demonstrated that visuomotor adaptation of grasping is not the primary source of illusion resistance in closed-loop grasping.

Keywords: Manual size estimation; Ponzo illusion; Reach-to-grasp; Visuomotor adaptation.

Figure 1. Geometry of the Ponzo illusion display and the starting hand position

Illusion is depicted with two physically equal-size objects to demonstrate perceived size distortion, but experimental trials included only one object. At bottom-left, neutral grid used during Baseline and Washout phases. At bottom-right, all five physical objects used during Baseline and Washout; only the middle three sizes were used during Illusion-Adaptation. The Perceptually Equalized phase involved two pairs of physical objects chosen from a larger set that were estimated from each participant’s slope-corrected MSE effect.

Figure 2. Experiment phases and design of trial bins for Illusion-Adaptation phase

The main experimental series involved three phases (Perceptually Equalized phase not shown). During the Illusion-Adaptation phase, each participant observed a trial sequence determined by randomly ordering six different trial bins (following one trial that primes the sequence, represented here by the offset leftmost square). These bins were constructed to allow balanced analysis of the previous-trial effect in each bin.

Figure 3. Quantitative predictions of the adaptation model

(A) Average simulated MGAs for illusory-large (dark red) and illusory-small (red) targets, computed from 2000 model simulations. Gray regions indicate when the Ponzo display is presented. There was no random error term in the simulated MGA response function—visible noise in these curves is due entirely to randomness in trial sequence generation and remains despite the large number of simulations. This figure was intended only as a demonstration of model behavior, not as an explicit prediction of the MGAs. We expected between-subjects variability and motor noise to obscure this pattern. (B) Predicted binned illusion effects. In the Baseline phase, the model predicted no difference in MSEs or MGAs between the left and right neutral display locations. In the Illusion-Adaptation phase, the model assumed a stable positive effect on MSEs (lines) and predicted an exponential decrement in the effect on MGAs (squares). In the Washout phase, the model predicted an anti-illusory aftereffect on MGAs. In the Perceptually Equalized phase, where trials alternated between illusory-large and illusory-small in a single location, the model predicted an enhanced illusion effect on MGAs. The proportional enhancement was approximately equal to the model’s error correction rate parameter. The parameters used for these simulations are reported in the Pilot Data. (C) The model also predicted a previous-trial effect in the first Illusion-Adaptation bin, where MGAs are larger when they follow a grasp to an illusory-small target than when they follow a grasp to an illusory-large target.

Figure 4. Pilot study results (N = 45)

The Ponzo illusion influenced both perception and grasping, and these effects were correlated, but only grasping seemed to become resistant to the illusion over 20 trials. Center: We found significant illusion effects on both manual size estimates (ΔMSE) and maximum grip apertures (ΔMGA). Left: Effects on perception and on grasping were positively correlated. Each dot is a participant. Participants who reliably reported the illusion via MSEs also tended to grasp according to the illusory sizes. Right: Evolution of ΔMSE and ΔMGA over twenty trials, shown alongside average results from the individual model fits. Top-right: The perceptual effect appears to have increased over the course of the experiment, perhaps because participants exaggerated or categorized their MSE responses with repeated trials. The model accurately estimated the mean perceptual effect from the MGA data. Bottom-right: Grasping was influenced by the illusion in the early trials but this effect gradually diminished over time (n.s.). The model, which was fit to each participant’s MGA time series by estimating size distortion and error correction rate parameters, roughly captures this pattern.

Figure 5. Previous-trial effect on MGA in early trials of the pilot study

Figure 6. Temporal analysis of illusion effects on MGA and MSE

(Top) Slope-corrected illusion effects by Bin. (Bottom) Average MGA and MSE for each combination of illusion context and physical size. Solid lines trace the Bin averages. Note: In the Perceptually Equalized phase (far right), targets in the illusory-small context were physically larger than targets in the illusory-large context by approximately 2 mm. This was accounted for in the illusion effect calculation.

Figure 7. Previous-trial effect on MGA

(Left) A significant effect was observed in Bin 1, as predicted by the adaptation model. (Right) The effect was also found when pooling data across the whole Illusion-Adaptation phase. A post-hoc test also revealed a previous-trial effect on MSE (not shown).

Figure 8. Post-hoc analysis of grip aperture trajectories in Illusion-Adaptation phase

(Top) Grip aperture trajectories for the illusory-large (left) and illusory-small (right) contexts. Dotted black lines are average Baseline grasp trajectories, specific to the location that the corresponding illusion context was presented. We used Baseline grasp data to estimate what the trajectory should look like for two “imagined” targets differing in physical size by the magnitude of the MSE effect (blue dashed line). If grasping were completely deceived by the illusion, the grip aperture trajectory would be expected to match this estimate until just before contact. Red lines are average grip aperture trajectories over the entire Illusion-Adaptation phase. Boxplots depict the spatial distribution of MGAs. (Middle) Separate illusion effect trajectories for illusory-large and illusory-small contexts, computed as the difference between the red and black lines depicted above. (Bottom) Combined illusion effect trajectory, computed as the difference between the separate illusion effect trajectories. This data evolves in line with the prediction until the hand reaches a point roughly 14 cm from contact, after which the effect is attenuated.

Figure 9. Visual summary: Reported temporal analyses of illusion effects on MGA and MSE

(A) Closed-loop experiments used the Ponzo illusion and measured both MGA (reds) and MSE (blues). Whitwell et al. (2016) measured the MSE effect before and after the grasping block, here we show the mean (light blue dashed line). MGA and MSE slopes were considerably less than 1 in Whitwell et al. (2016), so we slope-corrected the reported effects to improve comparability. Illusion effects were already slope-corrected in the present study. Our pilot study used only one target size so response slopes are unknown. (B) Open-loop experiments used the Müller-Lyer or Parallel Lines illusion and measured only MGA, not MSE. Illusion effects were already slope-corrected in Kopiske et al. (2017). The MGA slope was nearly 1 in both Franz et al. (2001) experiments. Note the difference in the ordinate scale between panels A and B.