Memory influences haptic perception of softness

Abstract

The memory of an object's property (e.g. its typical colour) can affect its visual perception. We investigated whether memory of the softness of every-day objects influences their haptic perception. We produced bipartite silicone rubber stimuli: one half of the stimuli was covered with a layer of an object (sponge, wood, tennis ball, foam ball); the other half was uncovered silicone. Participants were not aware of the partition. They first used their bare finger to stroke laterally over the covering layer to recognize the well-known object and then indented the other half of the stimulus with a probe to compare its softness to that of an uncovered silicone stimulus. Across four experiments with different methods we showed that silicon stimuli covered with a layer of rather hard objects (tennis ball and wood) were perceived harder than the same silicon stimuli when being covered with a layer of rather soft objects (sponge and foam ball), indicating that haptic perception of softness is affected by memory.

Figure 1

Setup and stimuli. (A) Setup of Experiments 1, 2 and 4. Participants sat in front of the experimenter. They made lateral strokes with the index finger to explore the surfaces of the stimuli and indented them then with a probe. A cardboard box with a blind prevented participants from seeing the stimuli. (B) Stimuli Experiment 1. Silicone rubber disks were partly covered with thin layers of bark (wood condition) or sponge glued onto elastic cloth. (C) Setup Experiment 3. Participants sat at a visuo-haptic workbench. Exploration of the stimuli was as in Experiments 1, 2 and 4. The probe was connected to a PHANToM force feedback device (for details about this setup see). (D) Tennis ball and sponge used for the covers in Experiments 2 and 4 and foam ball used for the cover in Experiment 3. (E) Stimuli in Experiments 2–4. Silicone rubber stimuli were covered with pieces of foam, or a tennis ball, or a slice of a sponge in Experiment 3. The foam ball cover was not used in Experiments 2 and 4.

Figure 2

Average PSEs and matches. (A) Experiment 1: Average perceived softness of an uncovered silicon standard measured with comparison stimuli covered with a layer of wood, sponge or uncovered. It is predicted that the standard will be perceived equal to harder comparisons when they are covered with the sponge and to softer comparisons when they are covered with the wood as compared to when they are uncovered. (B) Experiment 2: Average perceived softness of an uncovered silicon standard measured with comparison stimuli covered with a layer of a tennis ball, sponge or uncovered. It is predicted that the standard will be perceived equal to harder comparisons when they are covered with the sponge and to softer comparisons when they are covered with the tennis ball as compared to when they are uncovered. (C) Experiment 3: Average matches of a silicon rubber standard with a foam ball, a sponge, a tennis-ball cover or no cover to uncovered silicon rubber comparisons. It is predicted that the standard will be perceived harder when covered with the tennis ball and softer when covered with the sponge or foam ball as compared to when being uncovered. (D) Experiment 4: Average perceived softness of a silicon stimulus covered with a layer of a tennis ball, sponge or uncovered measured with uncovered silicon comparisons and differently long explorations of the stimuli (one touch or unlimited). It is predicted that the standard will be perceived harder when covered with the tennis ball and softer when covered with the sponge as compared to when being uncovered with effects being smaller with a long exploration. Error bars represent the within-subject standard error. *p < 0.05; **p < 0.005; ***p < 0.0005; ****p < 0.00005.

Figure 3

Average JNDs. Experiment 4: Average discrimination thresholds for softness of silicon stimuli explored with a pen given different length of the exploration (one touch or unlimited). JNDs were averaged over the object surface conditions. Error bars represent the within-subject standard error. **p < 0.005.