Eyes-Off Your Fingers: Gradual Surface Haptic Feedback Improves Eyes-Free Touchscreen Interaction

The Weber-Fechner law applies to the perception of the spatial frequency of real gratings.

“The Weber-Flechner law applies to the perception of the spatial frequency of real gratings [28,29]”

• cites [28] 28. Harold T Nefs, Astrid ML Kappers, and Jan J Koenderink. 2001. Amplitude and spatial-period discrimination in sinusoidal gratings by dynamic touch. Perception 30, 10 (2001), 1263-1274. sourceNot checked — no full text of the source
• cites [29] 29. Harold T Nefs, Astrid ML Kappers, and Jan J Koenderink. 2002. Frequency discrimination between and within line gratings by dynamic touch. Perception & psychophysics 64, 6 (2002), 969-980. sourceDisputed

  The General Discussion (page not provided) explicitly concludes that the Weber fraction is not constant for line-frequency discrimination. It reports that the fraction decreased as the frequency of the raised-line gratings increased within the experimental range, contrary to the claim that Weber’s law applies. (LLM, conf=0.88)

  Checked by a reasoning model — may be wrong.

  Experimental — accuracy not yet measured

Augmented feedback through visual, auditory or haptic modalities effectively enhances motor learning.

“Augmented feedback, through the use of visual, auditory or haptic modalities, has been shown to effectively enhances motor learning [38].”

• cites [38] 38. Roland Sigrist, Georg Rauter, Robert Riener, and Peter Wolf. 2013. Augmented visual, auditory, haptic, and multimodal feedback in motor learning: a review. Psychonomic bulletin & review 20, 1 (2013), 21-53. sourceSupported

The addition of haptic feedback can significantly reduce the time it takes to perform a task.

“Recently, it has been shown that the addition of haptic feedback can also significantly reduce the time it takes, effectively illustrating that the difficulty of the task is reduced in the presence of an additional type of feedback [12,20,23].”

• cites [12] 12. Géry Casiez, Nicolas Roussel, Romuald Vanbelleghem, and Frédéric Giraud. 2011. Surfpad: riding towards targets on a squeeze film effect. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems. 2491-2500. sourceNot checked — no full text of the source
• cites [20] 20. Farzan Kalantari, Edward Lank, Yosra Rekik, Laurent Grisoni, and Frédéric Giraud. 2018. Determining the haptic feedback position for optimizing the targeting performance on ultrasonic tactile displays. In 2018 IEEE Haptics Symposium (HAPTICS). IEEE, 204-209. sourceSupported
• cites [23] 23. Vincent Levesque, Louise Oram, Karon MacLean, Andy Cockburn, Nicholas D Marchuk, Dan Johnson, J Edward Colgate, and Michael A Peshkin. 2011. Enhanc- ing physicality in touch interaction with programmable friction. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems. 2481-2490. sourceNot checked — no full text of the source

Subjects are slower in the auditory learning conditions, even when the feedback disappears.

“Subjects appear to be slower in the auditory learning conditions, even when the feedback disappears.”

With haptic feedback only, subjects performed the task on average in 4.1 s with 8.1% of error.

“With haptic feedback only, subjects performed the task on average in 4.1 s with 8.1% of error (mean of the absolute value of the error and mean of the completion time on all subjects during the evaluation session, outlier trials excluded).”

After training with visual and auditory feedback, users are able to adjust a setting on a haptic touchscreen without looking at the screen.

“We demonstrate that, after training with visual and auditory feedback, users are able to adjust a setting on a haptic touchscreen without looking at the screen, thereby reducing visual distraction.”

• Also stated in: § 5.1 Overall results · plain

  “After training, participants were able to use the haptic cues to adjust the setting to γ t = 60 or γ t = 80.”

Task completion time is longer for conditions with audio feedback.

“Both the learning curves and the boxplot show that the task completion time is longer for conditions with audio feedback.”

After training, participants were able to use the haptic cues to adjust the setting to γ t = 60 or γ t = 80.

“After training, participants were able to use the haptic cues to adjust the setting to γ t = 60 or γ t = 80.”

The task distance shows a significant effect on the completion time (one-way ANOVA: F 6 = 12.8, p=3.3 • 10 -14 ).

“The task distance shows a significant effect on the completion time (one-way ANOVA: F 6 = 12.8, p=3.3 • 10 -14 ).”

In the training session, subjects with haptic-only and additional auditory feedback start with high error (a+c=17.1 and 12.6), while visual feedback subjects start much lower (a+c=5.8).

“Subjects with haptic feedback only and additional auditory feedback start with a high error (a + c = 17.1 and 12.6 respectively) whereas it is much lower for subjects with visual feedback (a + c = 5.8).”

Learning strategy has a significant effect (α=0.05) on absolute error during evaluation (one-way ANOVA F4=221, p=0.011).

“Statistical analysis exhibits a significant effect (at α = 0.05) of the learning strategy on the absolute value of the error (one-way ANOVA: F 4 = 221, p=0.011).”

Learning strategies have a significant effect (α=0.05) on completion time during evaluation (one-way ANOVA F4=13.66, p=5.9e-11).

“Learning strategies also have a significant effect (at α = 0.05) on the completion time during the evaluation session (one-way ANOVA: F 4 = 13.66,p=5.9 • 10 -11 ).”

In the particular interaction of tuning an integer value, surface haptics can significantly improve the interaction.

“Here, we show that in the particular interaction of tuning an integer value, surface haptics can significantly improve the interaction.”

The final error is approximately at the same level regardless of the training strategy.

“The transition to the evaluation session causes a significant gap for the visual and audio conditions, but the error finally appears to be approximately at the same level regardless of the training strategy.”

After training, subjects improved their setting performance, regardless of the learning strategies.

“We found that, after training, the subjects improved their setting performance, regardless of the learning strategies.”

The bias due to initial value is much lower than the theoretical bias with proprioception only.

“Here, even if the initial value shows a significant effect on the setting error (one-way ANOVA: F 4 = 19.9,p=5.4 • 10 -16 ), due to an overestimation for the γ i = 30 and 40 initial values, this bias is much lower than the theoretical bias with proprioception only.”

For fading-out feedback strategies, error increases mid-training when audio and visual stimuli become imperceptible, reaching the error level of the haptic-only condition.

“In contrast, for the learning strategies with feedback that fades out with time, the error increases at the middle of the training session when audio and visual stimuli become imperceptible, and reaches the error level of the haptics only condition.”

The initial value shows a significant effect on the setting error (one-way ANOVA: F 4 = 19.9, p=5.4 • 10 -16 ).

“Here, even if the initial value shows a significant effect on the setting error (one-way ANOVA: F 4 = 19.9,p=5.4 • 10 -16 ), due to an overestimation for the γ i = 30 and 40 initial values, this bias is much lower than the theoretical bias with proprioception only.”

There is a significant difference between the two target value conditions (t=-27.5, df=1591, p=3.8 • 10 -136 ).

“A two-sample Student-Welch t-test with unequal variances validates the significant difference between the two target value conditions (t=-27.5, df=1591, p=3.8 • 10 -136 ).”

The significant effect on error is due only to a slight difference between fading-out Audio and fading-out Visual conditions (Tukey p=0.012).

“This effect appears to be only due to a slight difference between the fading out Audio and fading out Visual condition (Post hoc Tukey's test: p f V -f A = 0.012)”

Every learning strategy yielded similar performance.

“Every learning strategy yielded similar performance, suggesting an amodal integration.”

• Also stated in: § 1 INTRODUCTION · plain

  “We found that, after training, the subjects improved their setting performance, regardless of the learning strategies.”

The transition to the evaluation session causes a significant gap for the visual and audio conditions.

“The transition to the evaluation session causes a significant gap for the visual and audio conditions, but the error finally appears to be approximately at the same level regardless of the training strategy.”

The effect on completion time is due to Visual, Audio, and fading-out Audio conditions leading to significantly higher completion time.

“The post hoc Tukey's test reveals that this effect is due to the Visual, Audio and fading out Audio conditions that lead to a significantly higher completion time.”

Subjects spent more time to adjust the setting when the target value was much greater than the initial value.

“Subjects spent more time to adjust the setting when the target value was much greater than the initial value, as exhibited by the linear regression performed on the mean completion time T = 0.035 × ∆ γ + 2.42 (R 2 = 0.993).”

Some participants reported focusing on haptic feedback and paying less attention to visual or auditory feedback.

“As a consequence, some of them reported that they had focused on the haptic feedback and paid less attention to the visual or auditory feedback.”

After even a short training session, differences in performance vanish.

“However, after even a short training session, the differences in performance vanishes, which is a promising result for enabling eyes-free control of sliders.”

Retention after training did not depend on the learning modality.

“Since retention after training did not depend on the learning modality, the learning method can be adapted to different contexts and to the users' preferences.”

For the same task with visual feedback (slider on screen), setting error was approximately 2.2% (mean of last 100 trials of H+V condition).

“In comparison, for the same task with visual feedback in the form of a slider on a screen, the setting error was approximately at 2.2% (mean of the last 100 trials of the H+V condition).”

Learning curves show a plateau after about 40 trials, or four blocks.

“Additionally, the learning curves show a plateau after about 40 trials, or four blocks, so in practice the training session could probably be significantly shortened.”

Completion time increases with larger distances, similarly to Fitts' law.

“The completion time increases with larger distances, similarly to Fitts' law.”

Subjects needed about 35 trials to reach 95% of the final error level with only haptic feedback.

“The learning curve during the training sessions with only haptic feedback demonstrates that subjects needed about 35 trials to reach 95% of the final error level, which shows that the interaction is mastered remarkably quickly.”

Subjects maintain slower exploratory motion even when audio feedback is removed, in the fading-out condition and evaluation.

“Remarkably, subjects maintain this slower exploratory motion even when the audio feedback is removed, for the fading out condition and for the evaluation.”

The relationship between distance and time is linear, not logarithmic.

“However, we found that the relationship between distance and time is linear and not logarithmic, perhaps due to the nature of the interaction, which differs from a pointing task and thereby involves other control mechanisms.”

• Also stated in: § 7 CONCLUSIONS · boosted

  “The mean completion time (with outliers exclusion) increases linearly with respect to the task distance, as proven by the linear regression presented in blue line (R 2 =0.99).”

Participants adjusted a setting on a 0-100 scale with an error of 8.1% using progressive haptic feedback only.

“Participants were able to adjust a setting on a 0 to 100 scale with an error of 8.1%, using a progressive haptic feedback only.”

• Also stated in: § 7 CONCLUSIONS · plain

  “After learning, users succeeded in performing the eyes-free task with 8.1% of error and in 4.1 s on average.”

The type of training feedback had little effect on final performance during the evaluation session.

“We found that the type of training feedback had little effect on the final performance during the evaluation session.”

Surface haptics can provide intuitive and precise tuning possibilities for tangible interfaces on touchscreens.

“This study shows that surface haptics can provide intuitive and precise tuning possibilities for tangible interfaces on touchscreens.”

The similar performance across learning strategies suggests an amodal integration.

“Every learning strategy yielded similar performance, suggesting an amodal integration.”

Subjects succeeded in integrating tactile cues in their selection.

“It demonstrates that subjects succeeded in integrating tactile cues in their selection.”

The sudden removal of a feedback can cause a significant degradation of performance.

“Indeed, the sudden removal of a feedback can cause a significant degradation of performance [19].”

• cites [19] 19. Felix C Huang, R Brent Gillespie, and Arthur D Kuo. 2006. Human adaptation to interaction forces in visuo-motor coordination. IEEE Transactions on Neural Systems and Rehabilitation Engineering 14, 3 (2006), 390-397. sourceNot checked — no full text of the source

To develop a persistent internal representation of the task, feedback frequency and intensity should be progressively decreased.

“However, to develop a persistent internal representation of the task, feedback frequency and intensity should be progressively decreased [22].”

• cites [22] 22. Attila J Kovacs and Charles H Shea. 2011. The learning of 90 continuous relative phase with and without Lissajous feedback: external and internally generated bimanual coordination. Acta psychologica 136, 3 (2011), 311-320. sourceNot checked — no full text of the source

The just-noticeable difference of vibration frequencies is between 18% and 20%.

“the just-noticeable difference of vibration frequencies (i.e. the lowest frequency difference to perceive a change in vibration) is between 18% and 20% [34]”

• cites [34] 34. Helena Pongrac. 2008. Vibrotactile perception: examining the coding of vibrations and the just noticeable difference under various conditions. Multimedia systems 13, 4 (2008), 297-307. sourceNot checked — no full text of the source

Binary feedback only guides the user when a specific target is reached and not during the entire motion.

“However, binary feedback only guides the user when a specific target is reached and not during the entire motion.”

The new guidance method, leveraging continuous mapping of haptic cues to an abstract value, has only marginal degradation of performance compared to visual cues.

“This new guidance method which leverages a continuous mapping of the haptic cues to an abstract value has only marginal degradation of performance compared to the use of visual cues.”

The task with only haptic feedback is initially cognitively demanding.

“The initial trials with only haptic feedback were two times more laborious than the steady state performance and about three to four times longer than with visual feedback, suggesting that the task with only haptic feedback is initially cognitively demanding.”

The absence of a larger effect could be due to a too small sample size, given large inter-subject variance.

“Since the inter-subject variance is large, the absence of a larger effect could be due to a too small sample size.”

By feeling a continuously evolving texture, control parameters can be adjusted in an absolute way, a task that until now seemed impossible without vision.

“By feeling a texture that evolves continuously, the control parameters of the interface can be adjusted in an absolute way, a task that until now seemed impossible without vision.”

• AssumptionRests on an assumption: The gradual haptic feedback is constructed based on the Weber-Fechner Law.

The gradual haptic feedback does not present any additional cues at the two target locations.

“However, these targets values were chosen arbitrarily and the gradual haptic feedback does not present any additional cues at these particular locations.”

Terminal feedback given on a digital display seems to be the essential factor enabling the retention process.

“Finally, given the relatively low impact of congruent feedback, we can conclude that the terminal feedback, given on a digital display, seems to be the essential factor enabling the retention process.”

The interaction is mastered remarkably quickly.

“The learning curve during the training sessions with only haptic feedback demonstrates that subjects needed about 35 trials to reach 95% of the final error level, which shows that the interaction is mastered remarkably quickly.”

A new control device allows setting a value without looking at it using a new haptic touchpad.

“We propose a new control device that allows setting a value without looking at it, using a new haptic touchpad.”