Twenty-eight healthy psychology students were equally allocated to one of two conditions. The bimanual grasping group (Experiment 1a, 4 males, average age = 24.7, SD = 2.9), and the bimanual perceptual estimations group (Experiment 1b, 5 males, average age = 25.1, SD = 1.4). The participants provided informed consent and received monetary compensation for their participation (the equivalent of 10$ for the longer grasping condition or 5$ for the shorter manual estimations condition). The results of one participant from the bimanual perceptual estimation group, who did not follow the experimental instructions, were removed from the analysis.
The experimental protocols in experiments 1 and 2 were approved by the Human Subjects Research Committee at Ben-Gurion University of the Negev (submission #1257). The study adhered to the ethical standards of the Declaration of Helsinki. All participants signed a consent form prior to their participation in the experiment. The manuscript contains no information or images that could lead to identification of a study participant.
Apparatus and stimuli
The participants’ grip scaling was tracked using an Optotrak Certus device (Northern Digital, Waterloo, ON). The apparatus tracked the 3D position of two active infra-red light-emitting diodes attached separately to the participant’s left and right index fingers (200 Hz sampling rate). We used Computer-controlled PLATO goggles (Translucent Technologies, Toronto, ON) with liquid–crystal shutter to control stimulus exposure time, and a custom-written MATLAB code to control trial sequences and events (version 9.4, The Mathworks, Natick, MA).
Stimuli were three rectangular-shaped plastic rods (21 cm, 22 cm, 24.5 cm in length, 1 cm in width and depth, for the small, medium, and large object, respectively). Lengths were chosen based on pilot experimentation in which we measured the physical difference between the objects that could generate an illusory effect in the opposite direction. In the illusory background condition (grasping and manual estimations), two objects were placed on the upper and lower sides of a standard or inverted version of the Ponzo illusion (Fig. 1). The purpose of using the inverted configuration was to allow proper control for measuring the effect of the illusion for objects located at the same distance from the participant. Without the inverted configuration, the illusory effect would always be confounded with distance from the observer, and distance from the observer has been shown to have a significant effect on grip apertures34. Therefore, to focus on the effect of the illusion, it is important to compare grip apertures between objects located at the same distance from the participant.
In the control grasping condition, the objects were placed against a white background. The objects were always placed at fixed locations. The distance between the initial location of the fingers and the nearer object was 24 cm, and between the two objects was 22.5 cm. The participants’ viewing distance was approximately 40 cm for the nearer object.
The procedure was based on the procedure used by Ganel et al.18 and was adapted for bimanual grasping control. In Experiment 1a, participants grasped objects placed on the illusory background or the control, non-illusory background. The two conditions were performed in separate blocks, and block order was counterbalanced between participants. Before each trial, participants placed their index fingers around a small circular disk (2 cm in diameter), used as the starting point, while the goggles were set on their non-transparent state. In the illusory background condition, the participant was asked to lift the object she perceived as the shorter/longer in the pair using both her hands. A verbal command signaled the required judgment (“short”/”long”), and an additional tone signaled movement initiation. In the control grasping condition, the participant was asked to lift the closer/farther object according to the corresponding verbal command (“close”/”far”). Each trial began with the verbal command which was followed by the openning of the goggles that stayed in their transparent state for 5 s, allowing on-line vision of the display and objects during the trial. The go signal was presented 1 s after the opening of the goggles.
The design of Experiment 1b was similar to the one used in Experiment 1a (illusory background condition), except that now the participants estimated the sizes of objects by manually adjusting the distance between their fingers following the opening of the goggles. Perceptual estimations were performed on the lower right side of the tabletop. To allow equal haptic feedback to the one provided in the grasping condition, following each estimation trial the participants were asked to grasp and lift the target object using both hands. Following perceptual estimates, a “go” signal to grasp the object was presented 5 s after the opening of the goggles. The goggles remained open for additional 3 s to provide online vision during grasp.
In each illusory background condition (grasping; perceptual estimations), a total of sixty experimental trials were used. The standard configuration was presented for thirty consecutive trials, and the inverted configuration was presented in the remaining trials. Presentation order was counterbalanced between participants. Overall, there were 44 critical incongruent trials, in which the physically longer object (22 cm) was placed in the illusory background that induced an illusory effect in the opposite direction (22 trials in each illusory configuration). There were eight additional incongruent catch trials in which the physically shorter object was placed together with a noticeably larger object, 24.5 cm in length. In this configuration, the illusory background did not lead to reversal of the perceived size differences. In the eight remaining catch trials, the smaller and the larger objects were placed in illusory congruent locations. Trial order was pseudorandomized. In the control grasping condition, there were a total of thirty-two trials; in half of the trials, the 21 cm object was closer to the participant, and in the remaining trials the 22 cm object was closer (presentation order was counterbalanced). The 24.5 cm object was not presented in the control condition. Each condition started with a few practice trials.
We recorded the 3D trajectories of the fingers in each trial. Movement onset was set as the point in time at which the aperture between two index fingers increased by more than 0.1 mm per frame for at least 75 ms. Movement offset was set as the point in time at which the aperture between the fingers changed less than 0.1 mm per frame for at least 125 ms (25 frames), but only after reaching the maximum grip aperture (MGA). The criteria used to detect movement onset and offset were based on the values used in a previous study from our lab in which we focused on unimanual grasps35. The values were slightly adjusted to allow effective identification of movement onset during bimanual grasping movements.
For our primary analysis, we calculated the average MGA for each object size. The MGA provides reliable measure of the sensitivity of the grip aperture to object size27,28. To further analyze the movement trajectory, we divided the raw movement data to 11 equal intervals and calculated the average aperture in each interval (see Online Appendix 1). Detailed data for each of the participants in experiments 1 and 2 is provided in Online Appendix 2.
To avoid a potential confounding effect of obstacle avoidance when reaching out to grasp the distant object, we focused our main analysis only on near objects. Previous research suggests that physical stimuli in the target’s environment can be treated as obstacles and affect kinematic aspects of the action toward the target37,38,39,40. For example, studies have shown that stimuli in the target surroundings could lead to deviations in movement trajectories from their original location39,40. As can be seen in Figs. 1 and 3, in the current study the near objects could be treated as obstacles, and could potentially affect the apertures between the fingers in flight. This would not have been a major issue if the obstacles were identical in size. However, they are not; when reaching out to grasp the physically far small object, the obstacle is always large (22 cm in Experiment 1) and when reaching out to grasp the physically large far object, the obstacle is always small (21 cm in Experiment 1). This potential confound makes it difficult to interpret the results of grasping objects in a far location. We therefore do not include the objects in this location in our main analysis and focus on the near location. Grasping the near objects is free of potential influences of obstacle avoidance and provides valid measure of grasping performance of objects located in a fixed distance from the participants’ starting position.
Thus, we compared the MGA obtained in two separate configurations of the illusion (standard/ inverted). In Experiment 1, the critical comparison was between the 21 cm object presented on the inverted configuration of the illusion (and perceived in most trials as larger) and the 22 cm object on the standard configuration of the illusion (perceived in most trials as smaller). We note that previous studies argued that such a between-blocks analysis could potentially prevent unwarranted effects of an attentional mismatch12.
In addition, as in Ganel et al.’s18 study, in order to disentangle the effects of apparent and real size on grip apertures, we focused our analysis on the subset of trials in which the participants erroneously judged the object as shorter/longer due to the illusion. In Experiment 1a (illusory background condition), the participants’ mean error rate for the incongruent trials was 80.9% (SD = 11.3, 35.2 trials on average). The error rate for near objects was 72% (SD = 15, 15.8 trials). The overall error rate in the perceptual estimation task was slightly larger overall (79%, SD = 16, 34.7 trials), as well as for the near objects (75%, SD = 17, 16.5). However, the difference in the overall error rate between the estimation and grasping conditions was not significant [t(24) = 0.3, p = 0.7], as well as for the near objects [t(24) = 0.3, p = 0.6].
Results and discussion
The main results are presented in Fig. 2. As can be seen in the figure, a dissociable pattern of performance was obtained for grasping grip apertures and for perceptual estimates of size. For grasping (Experiment 1a), the scaling of apertures reflected the actual size differences between the objects rather than their apparent sizes [t(13) = 4.2, p = 0.001]. The MGA was larger for an object that was perceived as smaller (253, 260 mm, for the short and large objects, respectively). In sharp contrast to grasping, manual estimations (Experiment 1b) went in the same direction as the overt perceptual judgments and were biased by the illusion. Therefore, the distance between the fingers during perceptual estimations reflected the illusory size difference between the objects [215, 205 mm, for the short and large objects, respectively, t(12) = 2.6, p = 0.02]. A mixed model ANOVA was conducted on grip apertures with experiment as a between-subjects independent variable and size as a within-subject factor. There was a significant main effect for experiment [F(1,25) = 94.6, p < 0.001, ηp2 = 0.80], with larger apertures for grasping (measured at MGA), compared to manual estimations (257, 210 mm, respectively). The main effect of size was not significant [F(1,25) = 0.92, p = 0.3]. Importantly, the interaction between physical size and experiment was significant [F(1,25) = 17.2, p < 0.001, ηp2 = 0.40].
The pattern of results in the control grasping condition in Experiment 1a was similar to that found in the main grasping condition with the illusory background. The MGA was smaller for the 21 cm object compared to the 22 cm object [260, 267 mm, t(13) = 6.0-, p < 0.001]. To look for potential differences between the illusion and control grasping condition, a repeated-measures ANOVA test with condition and size as within-subject variables was conducted on the MGA data. The test showed a main effect for condition [F(1,13) = 13.5, p = 0.003, ηp2 = 0.50], with larger apertures in the control condition compared to the illusion condition (263, 256 mm, respectively). The main effect of size was also significant [F(1,13) = 51.7, p < 0.001, ηp2 = 0.80]. The difference in overall MGA between the control condition and the illusory condition was not predicted and we do not have a solid account for this effect. More importantly, however, there was no interaction between condition and size [F(1,13) = 0.13, p = 0.7], indicating similar sensitivity to physical size differences in the two conditions. An analysis of the grip apertures across the movement trajectory in Experiment 1a is presented in Online Appendix 1. The pattern of results was similar to that obtained for the MGA data. Again, grip apertures reflected the real size differences between the objects. Mean MGAs and grip apertures in the manual estimation condition across the different distances from the target are presented in Table 1.
The results of experiments 1a and 1b extend and replicate previous findings obtained in the domain of unimanual control18. While bimanual estimations reflected the illusory size difference between the objects, bimanual grasping apertures reflected their actual difference in size. In Experiment 2, we tested whether this pattern of results extends to a different illusion of size, the Wundt–Jastrow illusion.