Keywords

1 Introduction

The development of vibrating tactors has focused on physiological characteristics when optimizing systems for human perception [1,2,3]. Neuropsychological research has focused on cognitive and neural correlates of tactile perception and memory [4]. An understanding of both the physiological and neurophysiological requirements is needed for the development of advanced human-in-the-loop, tactile based systems.

Vibrotactile cues can provide user information ranging from simple alerts for attention management (e.g., cell phone vibrations) to direction, spatial orientation, and more complex communications [5,6,7,8,9]. Quantitative meta-analyses of over 40 empirical studies showed significant positive impacts of tactile cueing on operational workload and performance, particularly when workload and attentional demands are high and/or when tactile cues are added to augment visual cues [5]. Complex vibrotactile cues driven from dynamic activation of multiple tactors have been developed to be intuitively understood, with little or no training [10,11,12]. Through these studies, tactile systems have demonstrated several key advantages when added to dismount Soldier navigation and/or communication systems, including human- robot interaction systems (e.g., communications from other Soldiers and also from robotic sensors).

In this report, we describe efforts to further investigate attributes of multi-tactor taction cues used to communicate a variety of alerting messages, with regard to perceptions of salience, ease of learning and recognition, and recall. We build upon previous investigations that found differences in response time and accuracy based on characteristics of tactors, such as amplitude and gain, and identified tactor engineering characteristics most likely to be perceived and recognized [13]. In this report we focus on additional aspects of multi-tactor taction differences, with regard to temporal sequencing and complexity of tactor signaling. Results described here are based on two studies, a preliminary exploration using 8 tactions, and a second study refined by results of the first experiment, that used 12 tactions.

Tactors and Tactile Belt.

For this effort, we utilized 2 types of tactors developed to optimize human tactile perception, the Engineering Acoustics Inc (EAI) C-3 tactor and the EAI EMR tactor, as shown in Fig. 1. The EMR produces about 0.7 mm displacement amplitude, with an operating frequency around 80–120 Hz. The EMR uses rotational motors that are suspended in a unique actuator configuration. The C-3 tactor is similar in performance to the C-2, but is lighter and has a smaller diameter. The C-3 utilizes a unique engineering approach proven to be particularly salient under strenuous movement [13, 14]. The contact with the skin is from the predominant moving mass, driving the skin with perpendicular sinusoidal movement that is independent of the loading on the housing [14]. The tactors are mounted in two rows (one row of 8 EMR tactors and one row of 8 C-3 tactors) within a belt form factor (Fig. 1).

Fig. 1.
figure 1

EAI EMR and C-3 tactor transducers (left to right), and 16-tactor tactile belt

Full size image

Tactions.

Tactions refer to the tactile patterns that are generated on a tactile array with characteristic features such as; spatial location, movement, temporal (pulse, rhythm and meter), frequency and intensity. Tactions are felt by the user, interpreted and associated with meaning. In the first study, eight tactions were created through using visual graphics software to easily create, save, and modify taction characteristics. A subset of the initial tactions were developed and used in previous studies using Soldier subjects and were found to be easy to perceive and interpret [15]. For the second study, four additional taction swere developed.

Tactions for our study were developed to vary systematically along two dimensions: (a) temporal sequencing and (b) complexity of tactor signaling characteristics. Regarding temporal sequencing, we followed Barber et al. [16] for the definition of static versus dynamic patterns. Static tactions present a constant pattern using the same tactors in a repetitive fashion. Dynamic tactions using used a sequenced presentation on different tactile locations that provides a sensation of movement across the tactors. We defined the level of complexity as Standard or Complex. Standard tactions use combinations of tone burst pulsating vibrotactile patterns as described in the review by Sarter and Jones [1]. These tone burst patterns are typically single frequency and can be pulse length modulated as described by Brewster and Brown [17]. Complex tactions use amplitude and/or frequency sweeps and/or short pulsatile sequences to create somatosensory illusion experiences (usually associated with the perception of movements). Examples would include various illusions such as the cutaneous rabbit [18], paint-brush illusion [19] and phi (motion; [20]).

We describe one example of each of the four kinds of tactions, using a screen shot of the taction creation software, which describes which tactor is activated. Tactors 1–8 are EMR tactors arranged sequentially in one row of the belt, tactors 9–16 describe C-3 tactors arranged similarly in a second row on the belt. Tactor 1 and 9 are thus located on the belly of the participant. In experiment 2, we used 12 tactions, four of each category.

Standard/Static.

Standard/static tactions were typified by static “pulsing” of tactors that did not vary in signal characteristics. They were simple and repetitive. As an example, the NBC (nuclear biological chemical threat) taction comprises a dynamic sequence of alternating pulses (between the back left front right). It was implemented on C-3, tactors 9–16, as portrayed in Fig. 2. Each tactor pulse in the sequence comprises a 250 Hz tone-burst at the maximum displacement.

Fig. 2.
figure 2

NBC “standard/static” taction using the C-3 (tactors 9–16)

Full size image

Standard/Dynamic.

Standard/dynamic tactions used standard tactor signal characteristics, with a dynamic temporal sequencing on multiple tactor locations. As an example, the adapted Rally taction comprises a dynamic sequence of pulses starting in the center (belly) and moving clockwise around the body that is repeated twice. It was implemented on C-3, tactions 9–16, as portrayed in Fig. 8. Note that this is a slightly different implementation of Rally from previous experiments; tactors were somewhat overlapping in duration (Fig. 3).

Fig. 3.
figure 3

Rally “standard/dynamic” taction utilizing a pulse sequence on the C-3 (tactors 9–16)

Full size image

Complex Static.

Complex/static tactions were “static” in the sense that the pulse stimuli were presented on fixed tactors in the belt array. However, each tactor pulse was “complex” in that the amplitude or gain was ramped linearly. As an example, the IED taction utilized 9 simultaneous pulses. It uses both the EMR and C-3 tactors in the taction. Specifically, each tactor was pulsed on for 1,500-ms tone-burst duration while the gain was linearly varied from maximum to zero (254-1 gain) (see Fig. 4). Thus, this tactions would be felt as an initial strong burst that “levels out” (e.g., IED).

Fig. 4.
figure 4

IED “complex/static” taction utilizing EMR (tactors 1–8) and C-3 (tactors 9–16)

Full size image

Complex Dynamic.

Complex/dynamic tactions used tactors with complex characteristics (i.e. “ramped” characteristics of gain or amplitude), along with dynamic temporal sequencing. For example, the Move Up taction comprises of sequenced ramps on two (or four) tactors in adjacent rows tactors. Thus, the C-3 and EMR tactors are used and ramped simultaneously. This taction pattern is dynamic following the tactile “paintbrush” illusion [19] and provides a sensation of back and forth movement over the front torso (Fig. 5).

Fig. 5.
figure 5

MoveUp “complex/dynamic” taction that utilizes both the EMR (tactors 1–8) and C-3 (tactors 9–16)

Full size image

2 Method

Overview of the Assessment.

While two experiments were conducted, we focus on the second, which was refined based on experiment 1 results. In experiment 1, participants easily learned eight tactions, allowing us to add 4 tactions in experiment 2, ensuring we had three tactions of each type (12 tactions). Individual tactions were also refined in response to Soldier feedback from experiment 1; for example, shortening the overall length of a particularly lengthy taction, to be more similar in length to the other tactions, and creating more distinction with regard to the Rally taction.

Twenty Soldiers assigned to operational and training units located at Fort Benning, GA participated in the five-day assessment. The average of age of the Soldiers was 27.25 years. Soldiers ranked from E-4 to E-5, with 70% of the Soldiers being E-4. 40% of Soldiers reported an average of 3.38 years for time in service, while 55% reported less than one year of time in service. Soldiers were informed of the nature and purpose of the investigation and given the opportunity to opt out with no repercussions. They provided responses to demographic questionnaires, were assigned a roster number, which corresponded to a counterbalanced experiment design where each Soldier (a) provided ratings of tactile salience on each taction, (b) experienced training on taction meaning, then participated in two performance trials, one where s/he was standing stationary, and one where s/he moved along a 2 × 4 beam placed on the floor in a slightly raised square pattern. After a three-hour break, each Soldier participated in two more performance trials, to explore any effects on recall of taction meanings.

Tactile Salience: Training and Measurement.

Each Soldier was given an oral and hands-on training of the tactile system. They donned the belt and wore headphones that emitted pink noise, to eliminate any audio cues associated with tactions. The experimenter activated each taction in turn, to give the recipient an overall familiarity of all tactions.

Soldiers were given the following instructions: “We will be presenting you with 12 different patterns of tactile signals. We will let you feel each of them first, to give you an idea of what each one feels like, and how they differ. Then, we will give you the signals one at a time, and ask you to give each one a rating, from one to five, that indicates how strongly, or easily, you think each one can be felt.” They were presented with a poster describing the 5 point scale for salience, ranging from 1 = weak, blurred, faint, vague to 5 = noticeable, distinct, strong, salient. A previous investigation established the reliability of this rating-based approach to measurement, compared to more traditional forced-choice methodology [21]. Each Soldier provided ratings of salience for each taction. Each taction was presented twice.

Taction Meaning: Training and Ease of Learning.

After ratings of salience were collected, each Soldier was trained on the meaning of each taction. The tactions were trained first in sets of three. Each set included all tactions of a particular type (e.g., standard static). Three tactions were presented, with meanings. The instructor would repeat each taction of this set, in random order, until the Soldier labeled each taction correctly, three times in a row. After all 12 tactions were presented to the Soldier in this way, the instructor would repeat each of the 12 tactions, in counterbalanced order, until the Soldier was able to correctly identify each taction three times in a row. The instructor documented the number of times each taction was repeated to achieve requisite performance.

Performance Trials.

After each Soldier was fully trained on the meaning of each taction, they participated in two performance trials in the morning, and two performance trials about three hours later. They did not get refresher training before the afternoon sessions and were asked not to discuss their experience with each other. An experimenter accompanied them during this break time. Sessions were counterbalanced according to Table 1.

Table 1. Assignment of soldiers to conditions. The stationary test condition refers to tactions being presented with the participant standing, and moving refers to one where s/he was required to move along a 2 × 4 beam placed on the floor in a slightly raised square pattern.
Full size table

3 Results

3.1 Salience

As the figure below shows, both main factors (Static vs. Dynamic, Standard vs. Complex) and the interaction term had main effects on ratings of salience. Repeated measures analyses of variance show a significant main effect for the static versus dynamic variable (F, 1, 19 = 6.67, p < 0.02, ηρ2 = 0.26) and for the standard versus complex variable (F 1, 19 = 56.18, p < 0.001, ηρ2 = 0.75) and for the interaction term (F 1, 19 = 97.37, p < .001, ηρ2 = 0.83l). Effect sizes as calculated by partial eta squared (ηρ2) were high. Results indicate that while static tactions were higher in salience, compared to dynamic tactions; this difference was significantly larger for complex tactions (Fig. 6).

Fig. 6.
figure 6

Salience: simple vs. complex tactions: static vs. dynamic

Full size image

3.2 Ease of Learning

Soldiers provided very high ratings of training effectiveness, ranging from means of 5.20 to 5.80 (7pt. scale). Each time the Soldier participant made an error, the instructor noted the error, and the taction that it was mistaken for, and communicated the correct taction meaning. This process was repeated, going through each of the twelve tactions, until the Soldier correctly identified each taction three times in a row. Tactions associated with highest total error during this learning process were “Target Detected” (standard/dynamic; 13 errors), “Move up” (complex/dynamic; 12 errors), “Wheel spin” (standard/dynamic; 5 errors), “Disperse” (complex/dynamic; 5 errors), and “Freeze” (5 errors). Aside from “Freeze”, these tactions were dynamic. The mean number of repetitions to learn the tactions to criterion performance ranged from 2.00 to 2.90. Some Soldiers learned all tactions very easily, while some had difficulty with most tactions, requiring four to five repetitions of all 12.

3.3 Performance and Recall

Figure 7 provides mean performance scores for each taction, by time of day. There were few differences in recall accuracy due to time of day; mean accuracy for tactions remained high, for tactions that were associated with high accuracy in the AM. There was some decline for standard dynamic tactions (Target Detected, Rally, Wheel spin). Repeated measures ANOVA examining three factors and interactions showed a significant effect due to Static vs. Dynamic factor (F 1, 19 = 30.16, p < 0.00, ηρ2 = 0.61), and a significant interaction between Standard vs Complex factor and AM vs. PM (F 1, 19 = 4.13, p = 0.05, ηρ2 = 0.13).

Fig. 7.
figure 7

Taction categories by AM_PM

Full size image

3.4 Performance and Movement

Figure 8 provides mean accuracy scores for each taction category, by participant movement test condition. The movement test conditions were counterbalanced against order. Values were quite similar across movement conditions, with the exception of Wheelspin, which differed from 87.5% (stationary) to 71.3% (balance beam). Repeated measures ANOVA showed a significant effect for Static vs. Dynamic factor (F 1, 19 = 37.74, p < 0.0001, ηρ2 = 0.66), but not for Standard vs. Complex (F 1, 19 = 0.33, p = 0.57, ηρ2 = 0.02), or Stationary vs. Movement (F 1, 19 = 2.41, p = 0.14, ηρ2 = 0.11). There was a significant interaction for Static vs. Dynamic and Stationary-Movement, showing that the difference in movement condition had an effect depending on whether the taction was Static vs. Dynamic (F 1, 19 = 6.65, P < 0.02, ηρ2 = 0.26). Other interactions were not significant.

Fig. 8.
figure 8

Mean accuracy scores for each taction category, by participant movement test condition.

Full size image

Subjective Feedback.

Soldier ratings were overall positive, reporting high ratings for system comfort and fit. Ratings also indicated the tactions were easy to learn and recognize. Mean ratings were generally high for ease of perceiving the tactions, in general, and while moving. When asked which tactions were easiest to learn and remember, they listed “Rally” (n = 16, standard/dynamic) and “Point Right” (n = 13, standard/static). Most difficult tactions were listed as “Target Detected” (n = 14, standard/dynamic), “Move Up” (n = 12, complex/dynamic) and “Wheel spin” (n = 10, standard/dynamic). Post-session ratings of “ease of recognizing each cue”, based on a 7pt. scale where 7 = “extremely easy to recognize” were highest for “Point Right” (mean = 7.00) and “Rally” (mean = 6.68). While Rally is a dynamic taction, and thus predicted to be less easily learned, it is also a taction that directly emulates the Army hand and arm signal for Rally, when the hand is raised overhead and is rotated in a circular motion. This “link” to an existing and familiar cue is likely more easily learned and remembered. A follow-up study will examine the effects of longer time durations between initial and subsequent performance.

The mean rating (7pt. scale) for “operational relevance” was relatively high (mean = 5.41, SD = 1.12). Soldiers indicated that the system, when developed to be combat ready (e.g., secure network to Army systems, rugged, etc.), would be useful for situations requiring noise disciple and when visibility is low (e.g., night operations, dense vegetation, smoke, etc.). Suggestions were offered regarding form, fit, power usage, and additional capabilities [21, 22].

4 Summary and Conclusions

Salience.

Consistent with results from Experiment 1, mean ratings were higher for standard tactions compared to complex, and higher for static tactions compared to dynamic. However, Experiment 2 results, based upon a more powerful experiment design, describe a significant interaction where complex tactions were more negatively affected by dynamic characteristics. The trend suggests that simple repetitive tactions are perceived as more salient.

Ease of Learning.

Averaging across taction categories, standard static tactions also appear to be easiest to learn, having the lowest rate of error (total = 7), followed by complex static (total = 10), standard dynamic (total = 19) and complex dynamic (total = 20). Results follow the same trend as for salience: results suggest standard static tactions are more easily learned.

Performance.

Overall mean accuracy, averaged over time and movement condition, showed that Soldiers learned 9 of 12 tactions with over 90% accuracy (92–99%). Soldiers had most difficulty with “Move Up” (complex/dynamic; 78%), “Wheel spin” (standard/dynamic), and “Target Detected” (standard/dynamic). ANOVA showed differences in performance due to the static vs dynamic factor were significant.

Recall.

Comparison of AM versus PM performance showed little overall degradation in performance. However, analyses of taction categories showed significant decline in performance for more dynamic tactions, and for an interaction effect, such that standard dynamic tactions were most negatively affected by time.

Movement.

Comparison of stationary and balance beam conditions also showed little overall degradation in performance, with the exception of standard dynamic tactions, such that standard dynamic tactions were significantly lower in the movement condition.

Results show consistent trends in favor of standard/static tactions, in terms of ease of perception (i.e., salience), learning, recognition, and recall. These results serve to better guide developers of tactile cueing displays, when developing tactions for non-directional alerts. Results also suggest the importance of familiarity of a taction, when the perceived taction could be “linked” to an existing concept. In this way, the Rally taction, while dynamic, was more easily learned and recalled, compared to other dynamic tactions. The rally taction directly emulated the circular Soldier hand and arm signal for “Rally”.

Results also showed that participants could easily learn up to twelve tactions in a relatively short period of time. Experiment 1, which used eight tactions, had very little variance in learning or performance. While more variance was associated with twelve tactions, there were some participants who easily learned the 12 tactions with little repetition and 100% accuracy in performance. These results will inform subsequent investigations of taction characteristics and individual differences.