Feedback Fundamentals: Interactions Flush With Frustration

Indifferent Attendant

The restroom fixture described previously and shown in Figure 3 can be dubbed the “indifferent attendant,” characterized by a device that seems to ignore an attempted interaction unless the correct triggering action is performed and registered. Users find this attendant infuriating as it offers no indication as to how close they are to performing interactions correctly and are abandoned to reason for themselves.

Consider the angry feeling that bubbles up in this person when, in a public bathroom, he finds himself waving his hands under the faucet, trying to elicit a few seconds of water from it in a futile rain dance of guessed-at mudras. (Braverman, 2011, p. 17)

OPEN IN VIEWER

Figure 3.

Left: uninformative feedback of an indifferent attendant; right: natural feedback conveying more information.

When a user reaches under the faucet to trigger the water, it indicates that the perceived affordances – the possible interactions between person and product inherent to the design – were correctly identified (Norman, 2007). In the absence of levers or handles, users familiar with those cues are drawn toward the touchless interactions those perceived affordances provide.

The problem for indifferent attendants seems to be that they provide either poor artificial feedback or none at all. This can lead to irritation and anxiety on the part of the user, as feedback serves to reassure the user that he or she is still in control. Continuous feedback is needed to convey that a device works as intended and that any attempted interactions have been processed and received (Norman, 2013).

Current automatic faucets tend to forego intentionally designed feedback entirely, discounting the natural feedback provided by signals such as flowing water. But water serves to provide positive natural feedback only at the back end of the interaction. What about feedback that guides user understanding before ultimate failures? Perceptive users may be able to hear the opening and closing of solenoids controlling the water and thereby reason they are interacting correctly, even in the absence of designed feedback. However, most of the time the feedback is woefully inadequate for most users.

The first of Norman’s (2007) six design rules calls for the provision of “rich, complex, and natural signals” (p. 152). Although one might argue that there is little opportunity to provide useful feedback for such a simple device, Gaver (1997) might argue that the incorporation of natural sounds may leave users less frustrated. If instead of a beep or a click the user was fed the sound of gurgling water, the user could immediately know that, even though the device is trying, it is failing if the water doesn’t turn on.

A manufactured natural-sounding gurgle would also play nicely with Weiser and Braun’s (1995) concept of calm technology, where the feedback from an interaction is primarily in the user’s periphery. When the faucet functions as expected, the natural and artificial sounds combine, and the user is not distracted by the feedback. When the faucet fails to turn on, the presence of the artificial “natural” signal serves to assure the user that they have attempted the correct interaction and the user is free to move on to the next available faucet.

Restroom interactions are often controlled by infrared sensors and, depending on the quality of the algorithm, can be susceptible to interference from outside sources of infrared light, such as sunlight and incandescent lighting. More innovative integrated solutions that seek control over larger areas of restroom real estate can ensure much greater reliability, but at much greater cost. This creates difficulties for fixtures indifferent to providing good feedback.

During development cycles, the first author has observed strobe lights, lights triggered by motion sensors, and sunlight all fooling faucet algorithms to continuously trigger and waste water. Although all these failures are lessons for continued improvement, many occur under situations that are not easily replicated. In many failures, the root cause is not easily predicted by the designer or understood by the user. As a result, users and designers could be well served if the trigger conditions of a sensor were simply and clearly represented. During product development and installation, display representations could afford designers quick evaluation of algorithm revisions and display negative interferences. During implementation, feedback could afford users visible cues on the state and trigger condition of the device.

Passive-Aggressive Attendant

After lathering your hands, you place them under the faucet to rinse. This time the attendant obliges without hesitation, and water flows. A short moment into your handwashing ritual, however, the stream is abruptly interrupted. Although you attempt to restart it quickly, you are instead confronted by a moment of deliberate technological pause. “Shame on you,” the faucet seems to say with a curt, impatient click. “Think of the environment.”

Users of the purported “water efficient” infrared taps were observed experiencing problems with operation and their opinions were very negative. (Hills & Birks, 2004, p. 23)

Persuasive technology attempts to change attitudes or behaviors of users without coercion (Fogg, 2002; Harjumaa & Oinas-kukkonen, 2009). Some would argue the faucet uses persuasive techniques to change user behavior and aid water conservation. Not only is this frustrating for users, due to the lack of feedback from the faucet attendant, but it is a poor implementation of persuasive technology as it is coercive in nature. The motivation to conserve water cannot be expected to exist for the user; and the user is not persuaded to conserve, as much as he or she is forced to, leading to negative user opinions (Hills & Birks, 2004).

If limiting water flow to short intervals is truly motivated by conservation, and is not a failsafe designed to prevent faucets from running continuously on false triggers, perhaps less frustrating methods exist. Srivastava and Shu (2013) have shown that discretization, the division of resources into discrete units, reduces consumption (see Figure 4 for example). Perhaps water-efficient faucets could include consumption feedback in terms of discrete units that appear to overflow once the suggested limit is reached.

OPEN IN VIEWER

Figure 4.

Faucets with discretization. Left: underuse; right: overuse.

Tactless Attendant

In the privacy of the commode, your seat suddenly transforms from a harmless fixture into an uncouth bidet, as the electronic attendant unceremoniously conjures forth a frothing tempest within the receptacle below. Your reactionary shuffle or hop attempts to avoid the inevitable spritzing to no avail. Your humiliating dance does nothing to pacify the technological demigod.

The “tactless attendant” can be found hidden within the unholy autoflush features of modern urinals and toilets. Intended to save users from flushing manually, the tactless attendant commonly fails by either flushing too soon or not flushing at all. In a failure of the latter variety, users are bestowed with the unfortunate vestiges of former guests and are subsequently guilted into waiting to ensure they do not leave behind the same evidence.

An early flush can be equally horrific. Amid the expectation of quick relief, perhaps already delicately poised to minimize any unnecessary contact, you find yourself rudely interrupted and greeted by an unsavory bouquet of noise, sensations, and smells; often triggered at the slightest improper movement.

Some electronic eyes do not “notice” our desperate hands reaching out for water, for example, or suddenly “decide” to constantly flush, rinse, or dry, with no way of stopping their operation. (Braverman, 2011, p. 23)

Once again, the lack of useful feedback is largely to blame. There is a mismatch between the user’s understanding of the automatic flushing function and the fixture’s programmed reality. The user must be aware of the spatial constraints of the triggering action – how far he or she can move – to avoid a premature flush.

In this case, representations could be simple. Feedback that communicates, “I flush when you stand. And I think you’re sitting,” as shown in Figure 5 would do the trick. However, communicating that effectively and universally, to a diverse user base, can be difficult. Perhaps a simple indicator that lights up once the user has triggered the sensor is sufficient; the implied message being that once the user moves out of range, the fixture flushes.

OPEN IN VIEWER

Figure 5.

The tactless attendant. Left: no feedback; right: seated and standing display.

Untimely Attendant

While trying to receive soap, you may find that there is no reaction from your initial gesture. However, as you withdraw your hands, the soap suddenly dispenses onto the countertop; resulting in bemusement at the unexpected asynchrony. Often, the wasted material pooling on the countertop serves as the only source of informative feedback. These frustrated interactions occur with “untimely” technological attendants.

Whatever internal calculations the attendant needed to perform to complete the interaction, progress thereof was not appropriately communicated. Instead of informing users that gears were grinding, pumps were priming, or fluid was flowing, the attendant remained still, only to spring into life after the opportune moment had passed.

Another iteration of the untimely attendant is found in what is described as “reverse operation.” Extremely reflective surfaces beneath sensors create conditions in which it is possible for more infrared light to be returned to the sensor in the absence of a hand. Once hands are withdrawn, the signal is misinterpreted and soap is dispensed precisely when there is nothing waiting to receive it.

The lack of appropriate feedback during the initial failed interaction suggests to users that the device is not working, and so the rational decision is to try again or move on. Perhaps a series of indicators along the vertical height of the dispenser, lighting up successively and leading down to the outlet, would suffice (Figure 6). Such a display could afford users simple explanations for progress and the internal state of the device, rather than creating false assumptions.

OPEN IN VIEWER

Figure 6.

The untimely attendant. Left: no feedback; right: progress indicators.

Underperforming Attendant

Hands soiled after topping up your engine oil, you enter the service station restroom in search of hot water and grease-cutting soap. Unfortunately, the automatic attendant that you encounter is limited to binary control of a lukewarm drizzle. Instead of a quick moment cleaning up, you are forced to spend several minutes lathering and relathering your hands.

The above scenario is an example of a mismatch between assigned tasks and real abilities of a device that unfortunately results in reduced usability. Automatic faucets cannot provide users with the same level of control as manual faucets. This mismatch characterizes “underperforming” attendants (Figure 7).

These water-conserving devices are relatively dumb and can severely limit the functionality of existing sinks. They rarely allow users to decide the temperature or pressure of water, and for this reason are limited to public bathroom sinks. (Arroyo, Bonanni, & Selker, 2005, p. 632)

OPEN IN VIEWER

Figure 7.

The underperforming attendant, without means to adjust temperature.

Although the modern technological attendant was intended to fully replace the inconvenience of manual operation and ensure a more sanitary future, instead it settles for a compromise. In return for the novelty of touchless interaction, it wrests away control over both flow rate and temperature. There currently exist few intuitive ways to offer these controls for automatic faucets. Perhaps all that is missing is the proper metaphor and the establishment of a new convention. Representing and controlling temperature and flow rate is a relatively explicit endeavor. Could intuitive feedforward and feedback be designed for these controls for everyday people?

Several types of aftermarket aerators illuminate water in various colors according to faucet temperature. Perhaps it would be possible to set the temperature of automatic faucets through gestures over the faucet head. Feedback on the physical faucet could show the set temperature, and the color of water could show the temperature of the water. Users would therefore have control and feedback for the current set point as well as feedback for the current temperature. Whether such a solution could be intuitive enough for public use is speculative.

Embarrassingly, elements of the underperforming attendant appeared for the first author during development cycles of touchless dispenser models many years ago. Although the infrared sensors performed well in the lab, uncontrolled variables associated with location and lighting in the real world occasionally combined to cause the fixture to ignore dark hands. This resulted not only in an automatic dispenser that wouldn’t react for the dirtiest hands but also one that was unintentionally prejudiced against darker skin. By going to the trouble of automating dispensers, this relatively simple mechanical device had become not only significantly less usable but shockingly discriminatory and provided a sobering lesson on the merits of representative user testing.

Industry Solutions

There is little motivation for increased usability of restroom fixtures when it comes at the expense of reduced manufacturing competitiveness, regardless of the importance of good personal hygiene in combating the spread of infectious disease (Sax et al., 2007). So, is the everyday user forever stuck with poor restroom experiences that confuse and annoy?

There is reason to be optimistic. There are persuasive designs that attempt to elicit sustainable behavior (Arroyo, Bonanni, & Selker, 2005), and it’s not hard to imagine that there will soon be persuasive fixtures and faucets in industries, such as health care, that attempt to guide professionals more comprehensively through their hygiene rituals. Hand hygiene compliance for medical professionals still hovers only between 30% and 70% for the best Western hospitals (Reichardt et al., 2013). There is much room for improvement, along with the necessary motivation to do so (Klevens et al., 2007).

Although faucets and fixtures currently lack the means to provide useful feedback for public users, once these persuasive designs are more fully implemented in industry, designers for public spaces are afforded the necessary hardware to experiment with solutions that can create improved experiences and greater usability.

The public restroom lies firmly at the intersection of several factors that include public health, sustainability, and cross-cultural considerations. In navigating these problems, Moray (1995) argued that the challenge for ergonomics is to use social and cultural understanding and an emphasis on human needs, rather than wants, to affect persuasive behavioral change. As the demand for global solutions to hygiene challenges and resource conservation grows, restroom usability can stand to benefit from this elevated importance.

Applications

In discussing human-centered design issues, the modern restroom experience provides a convenient common denominator that is relatable and understandable to most people. There are several factors that uniquely complicate the public restroom space, as shown in the sidebar, however, the attendants discussed in this article are applicable in any discussion that involves interactive design and automation.

Applications include parking machines, ticketing kiosks, automated banking machines, vending machines, elevator controls, and hotel room lighting and temperature controls. Currently, most public technological interactions occur primarily through tactile means, but as the trend for gesture controls increases, it is conceivable that the future will be touchless.

The indifferent attendant leaves users without a basis from which to continue interaction. Once users have exhausted the incorrect affordances they perceive, the lack of feedback prevents them from progressing further. Any public technological attendant needs to provide feedback so that user misunderstanding can be corrected.

The passive-aggressive attendant forces unseen constraints and assumptions that run counter to users’ mental models. Public technological attendants should avoid having goals that disrupt or modify the user’s goals unless the user is sufficiently informed through feedback.

The tactless attendant makes assumptions about the state of the user and does not seek to communicate with, or adapt to, the user during the dynamic interaction. In order to facilitate understanding, public technological attendants should communicate with users at which stage of the interaction they find themselves.

The untimely attendant is slow to react to user input and leaves the user frustrated and irritated. Users are accustomed to immediate feedback, and in cases where that is impossible, indications of progress should be provided.

Finally, the underperforming attendant is charged with doing a manual task that it cannot fully replace. Public technological attendants should be designed to be equally or more usable than manual operation to avoid user frustration.