Illusory Motion Reversal in Touch

Abstract

Psychophysical visual experiments have shown illusory motion reversal (IMR), in which the perceived direction of motion is the opposite of its actual direction. The tactile form of this illusion has also been reported. However, it remains unclear which stimulus characteristics affect the magnitude of IMR. We closely examined the effect of stimulus characteristics on IMR by presenting moving sinusoid gratings and random-dot patterns to 10 participants' fingerpads at different spatial periods, speeds, and indentation depths. All participants perceived a motion direction opposite to the veridical direction some of the time. The illusion was more prevalent at spatial periods of 1 and 2 mm and at extreme speeds of 20 and 320 mm/s. We observed stronger IMR for gratings and much weaker IMR for a random-dot pattern, indicating that edge orientation might be a major contributor to this illusion. These results show that the optimal parameters for IMR are consistent with the characteristics of motion-selective neurons in the somatosensory cortex, as most of these neurons are also orientation-selective. We speculate that these neurons could be the neural substrate that accounts for tactile IMR.

Keywords: illusion; perception; perceptual rivalry; somatosensory; touch.

FIGURE 1

Experimental apparatus. (A) The miniature tactile stimulator with three motors, each of which controls one degree of freedom, including (a) rotation speed, (b) indentation depth, and (c) moving direction of the ball. (B) Three grating balls, with spatial periods of 1, 2, and 4 mm. The fingerpad was positioned palmar side up. (C) Motion stimuli were delivered to the fingerpad of the left middle finger via the stimulator. The red axes show the coordinates used for stimulus presentation and participant report. (D) For each trial, the motion stimulus was delivered for 1 s, and then the participant reported the perceived direction of motion by clicking the mouse on a circle on the computer display. During the experiment, the stimulator was housed inside a specially designed case, such that the participant could not see the tactile stimuli.

FIGURE 2

Analysis of IMR from distribution of perceptual bias – difference between perceived direction and veridical direction. (A) Two simulation distributions of perceptual bias generated by bimodal von Mises function. A₁ and A₂ reflect the amplitude of the highest peak and the opposite direction of the highest peak, respectively. A₁′ and A₂′ were computed by subtracting A₁ and A₂ with the minimal bin height (MBH) and IMR ratio = A₂′/A₁′. Data in the upper panel show obvious IMR, but data in the lower panel show weak IMR. (B) Single-participant responses from two different spatio-temporal conditions (upper panel: spatial period = 2 mm, speed = 80 mm/s; lower panel: spatial period = 4 mm, speed = 80 mm/s). The red curve is the bimodal von Mises fit for the binned perceptual bias (bin size = 15°). Results in the upper and lower panels had higher (IMR ratio = 0.052) and lower IMR (IMR ratio = 0), respectively.

FIGURE 3

Experiment 1: Spatio-temporal dependence of IMR. (A) IMR ratios across spatio-temporal parameters for one participant. (B) IMR ratio averaged across participants as a function of spatial period and speed (red: 1 mm, green: 2 mm, blue: 4 mm). (C) IMR ratio as a function of grating spatial period. (D) FWHM of A₁ averaged across participants as a function of spatial period and speed (red: 1 mm, green: 2 mm, blue: 4 mm). Bars indicate standard error of the mean.

FIGURE 4

Dependence of IMR on stimulus direction. We analyzed the distribution of stimulus directions that were putatively assigned to the IMR, by counting the trials with stimulus direction within range delimited by the range delimited by 180° perceptual bias ± 45°. Trial counts of IMR as a function of stimulus direction from each spatio-temporal combination. IMR tended to occur in certain stimulus directions (around 120° and –120°) and spatio-temporal combinations.

FIGURE 5

Experiment 2: Effect of indentation depth on IMR. The experimental paradigm was almost identical to Experiment 1, except that we presented speeds only in the middle-speed ranges and indentation depth as 250 or 500 μm. (A) IMR ratio under spatio-temporal (spatial period; red: 1 mm, green: 2 mm, blue: 4 mm) and indentation depth (solid lines 250 μm, dashed lines 500 μm) manipulations. (B) IMR ratio as a function of indentation depth showed that IMR ratio did not significantly depend on indentation depth (null hypothesis test F_(1,162) = 1.855, p = 0.175). The equivalence test also confirmed that there was no significant difference in the IMR ratio between indentation depths (Welch two sampled TOST equivalence test, boundaries = ±0.1886, DF = 143.14, p < 0.001). Bars indicate standard error of the mean.

FIGURE 6

Experiment 3: Orientation dependence on IMR. To examine whether orientation information is necessary to induce IMR, we presented a random-dot pattern ball. (A) Experimental apparatus. The random-dot pattern ball had dots randomly arranged on the ball surface with the average dot-to-dot distance of 3 ± 1 mm. The miniature ball stimulator and hand position were the same as those in Experiments 1 and 2. (B) IMR ratios across different ball types, including spatial period, speed, and indentation depth combinations. (C) The random-dot pattern induced the lowest IMR ratio compared with the grating balls for the spatial periods of 1 and 2 mm (1 mm vs. random-dot, F_(1,216) = 15.125, p < 0.001; 2 mm vs. random-dot, F_(1,216) = 13.530, p < 0.01) but not for gratings with spatial period of 4 mm (4 mm vs. random-dot, F_(1,216) = 0.049, p = 1). (D) FWHM across different ball types, including spatial period, speed, and indentation depth combinations. Bars indicate standard error of the mean.