Keywords

1 Introduction

Nowadays, the interaction with a computer system is common for most people. Latencies, defined as time delays between user input and system output, due to hardware and software characteristics are nearly unavoidable in human-computer-interaction [1]. If the latency of a system exceeds a certain threshold, users might get aware of the latency [2], the user experience can be impaired [3] and the user satisfaction can be reduced [4]. Different system, task and user related characteristics can influence the detection of system latency [5], but more research is necessary in this regard [6]. In the present study user related (i.e., motivation and visual processing speed) and task related influencing factors (i.e., movement type) on latency perception were examined.

1.1 Related Work

For the development of human-computer-systems it is important to know which latencies are perceivable by the user and impair the human-computer-interaction. Common design guidelines postulate a maximum latency of 100 ms to be sufficient to reach an optimal human-computer-interaction [7]. However, recent research could show that latencies considerably lower than 100 ms can be detected by the users (e.g., 64 ms; [2], 50 ms; [8], 60 ms; [6]). Results also indicate that the latency perception threshold differs dependent on several system related characteristics (e.g., input devices; [9]), task related characteristics (e.g., task complexity; [8]) or user related characteristics (e.g., experience with highly dynamic real-time computer games; [6]). There are some references in literature that also users’ motivation to identify latencies as well as users’ individual visual processing speed might be relevant user related characteristics [5].

In the present study the task related characteristic movement type representing one facet of task complexity and the two user related characteristics motivation and visual processing speed were examined.

Movement Type.

Research could show that it is more difficult for users to identify latencies in more complex tasks (e.g., writing with a digital stylus) than in less complex tasks (e.g., simple dragging tasks) [8, 10]. The authors postulated that more complex movements during task processing leads to a higher cognitive load and, hence, to less available cognitive resources for recognizing the system latencies [10]. Movement complexity can be defined by the Index of Movement Difficulty (ID) which can be calculated by the Fitts’ Law for straight movements [11] and by the Steering Law for circular movements [12]. Based on the research of [8, 10] it can be assumed that the latency perception is better for straight movement tasks (i.e., less complex movements) compared to circular movement tasks (i.e., more complex movements).

However, there is also research indicating that higher movement amplitudes (e.g., circular movements) could enhance the latency perception by providing more occasions to compare the input and the output signal [13] which should enhance the latency perception. One objective of the present research is to examine which task characteristic (i.e., movement difficulty or movement amplitude) are more prominent for latency perception in circular movement tasks compared to straight movement tasks.

Motivation.

Participants of the present research revealed the information that their individual latency perception threshold will be examined representing a performance test situation. Motivation in such a context can be defined as the willingness to invest effort to solve the task as good as possible over the whole time [14]. Research could show that participants’ motivation has an influence for instance on the performance in a simple detection task to examine participants’ perception threshold [15].

It can be assumed that all participants of the present study had a certain level of intrinsic motivation because they took part voluntarily in the experiment. However, external incentives (e.g., positive materially incentives) can also influence the performance [16]. Hence, it is assumed that a financial incentive increases participants’ external motivation to detect lower latencies and reduce the latency perception threshold.

Visual Processing Speed.

The perception process includes the processing of information [17]. How long it takes to process all the relevant information depends on the individual processing speed [18]. The processing speed for visual tasks (i.e., visual processing speed] can be examined for instance by using the digit-symbol-test of the Wechsler-Adult-Intelligence-Scale III [19]. Several research could reveal that more economic tests like simple or choice reaction time tasks [20] show medium to strong correlations with the digit-symbol-test and, hence, can be used as indicators for visual processing speed. Moreover, it is assumed that there is a strong correlation between the visual processing speed and the inspection time, defined as the minimal duration of a visual stimulus that produces a certain amount of correct answers [21]. It is assumed that the visual processing speed influences the latency perception threshold.

1.2 Hypotheses

The hypotheses examined in the present research are the following:

H1: The latency perception threshold differs between the more complex circular movement task and the less complex straight movement task.

H2: The latency perception threshold of the experimental group who received further external motivation through financial incentives should be lower than the latency perception threshold of the control group.

H3: Higher visual processing speed (examined with simple and choice reaction time tasks and inspection time tasks) correlates with a lower latency perception threshold.

2 Method

The present study was a 4x2 mixed measures design with the independent variables movement type and visual procession speed as within subject factor and motivation as between subject factor. The dependent variable was the latency perception threshold.

2.1 Participants

30 participants (67% female, Mage = 22.27) took part in the experiment. At the beginning of the experiment all participants signed an informed consent, including all information about the experiment and the anonymization of collected data.

2.2 Material

Software.

The experimental task was implemented in C++ by using the open source libraries of SFML [22] for 2D-displays in Code::Blocks (version 16.1.0). The technical system latency of the program was between 7.6 and 8.6 ms.

Hardware.

A computer with a 6-core-processor and a basic clock frequency of 3.3 GHz was used. The input device was a gaming mouse (Logitech G303) with 800 dpi. The tasks were presented on an Acer XF240H monitor with an update rate of 144 Hz.

System.

The mouse based 2D-dragging task included the navigation of a black square representing the mouse cursor through a white tunnel (straight or circular) as fast and accurate as possible (Fig. 1). Each trial consists of 2 sub trials with identical movement tasks but differing in the presented latencies. One sub trial had the technical system latency (i.e., no added latency) and the other sub trial has an added latency between one and 300 ms. After experiencing both sub trials in a randomized order the participants had to identify the sub trial in which the system reacts more instantly to the user input (i.e., the system without added latency).

Fig. 1.
figure 1

Movement types – straight (left) and circular (right). The blue arrows represent the movement direction of the mouse cursor and were not visible for the participants. (Color figure online)

Full size image

Examining the Latency Perception Threshold.

In every dragging-task an adaptive threshold estimation method (ZEST, [23]) was used to estimate the latency perception threshold based on the participant’s performance in the previous forced-choice-discrimination-task. With every correct (or false) answer of the participant the estimated perception threshold decreases (or increases) and, hence, the presented added latency of the next trial. The final latency perception threshold is defined as the added latency that can be correctly identified by the participant with a hit rate of 75% (see also [8]).

Movement Type.

The straight movement task (left side of Fig. 1) had an ID of 6.76 and the circular movement task (right side of Fig. 1) had an ID of 8.97. Participants had to move the mouse cursor once per sub trial from the starting point to the target area (represented by the two grey bars).

Motivation.

The experimental group received feedback regarding their latency perception threshold after the first part of the experiment. They were told that they will receive a financial incentive if they are able to improve their performance in the second part of the experiment. After each part of the experiment the participants filled out six items regarding their motivation on a 6-point-likert-scale (1 – fully disagree, 6 – fully agree).

Visual Processing Speed.

The inspection time task as well as the simple and choice reaction time task were used [20]. The inspection time task was implemented in Presentation® (version 19.0, www.neurobs.com). The same stimulus (two lines with different length) was presented to the participants several times with different durations. The participants had to correctly identify the shorter line. During the simple reaction time task the participants’ reaction time (button press) to a presented stimulus was examined. During the choice reaction time task participants, additionally, had to press one of four different buttons matching to the location on which the stimulus was presented.

2.3 Procedure

After a written instruction to the experiment as well as explanation of the term latency the participants started with a training in which they practiced the correct movement (i.e., straight and circular tunnel) as well as the latency perception task. After that, participants elaborated 30 trials for each movement type, followed by one test regarding participants’ visual processing speed. After a short break the experimental group received the invention and all participants elaborated again 30 trials for each movement type, followed by the other test regarding participants’ visual processing speed. After the experiment every participant was debriefed and received the financial incentive.

3 Results

Data analyses were done with IBM SPSS Statistics (Version 21.0) on a 5% significance level. Effect sizes were interpreted regarding [24].

3.1 Movement Type

Two outliers were identified regarding the conventions of [25] and were excluded from the analysis. Data revealed a difference between both movement types: Participants perceived lower latencies in the circular (M = 62.04 ms; SD = 29.24) compared to the straight movement task (M = 94.71 ms; SD = 59.49). A t-test for dependent samples revealed a medium effect: t(27) = –3.79, p = .001, d = –.59. Hence, H1 can be accepted.

3.2 Motivation

A repeated measures ANOVA revealed no differences between the experimental and the control group regarding the latency perception threshold (F(1, 28) = 0.336, p = .567). The manipulation check revealed, that both groups did not differ regarding their motivation after the intervention (t(28) = –0.96, p = .346, d = –.35), but were strongly motivated with M = 5.32 (SD = .51) for the experimental group and M = 5.10 (SD = .71) for the control group. Because the intervention regarding participants’ external motivation did not work in the present study, a linear regression with motivation as predictor was calculated over all participants. Results indicated that motivation is no significant predictor (\( \beta \) = –.099, t = –1.084, p = .281). Hence, H2 has to be rejected.

3.3 Visual Processing Speed

A Pearson correlation was calculated and revealed significant correlations between the values of the simple and choice reaction time as well as the inspection time task with the lowest identified latency perception threshold. Additionally, a multiple regression analysis was calculated with the values of the three tests as predictors. 41,9% of the variance could be explained (R2= .419, strong effect with f = .849). The choice reaction time task was the strongest predictor for the latency perception threshold and reached statistical significance (\( \beta \) = 393, p = .040). Hence, H3 is supported by the data.

4 Discussion

4.1 Movement Type

The results of the present study revealed that a slightly higher ID has no negative influence on latency perception opposed to the results of [8, 10]. Indeed, the more complex, circular movement leads to a lower latency perception threshold. These results are in line with the research of [13] and contribute to the assumption that higher movement amplitudes could enhance the latency perception by providing more occasions to compare the input and the output signal. Possibly, this effect overlapped the assumed negative impact of the slightly higher task complexity.

More research should be done with more complex movement types (e.g., several curves) to better understand the relationship of task complexity and movement amplitudes. Furthermore, in the present study movement type and ID were manipulated through the same variable (straight vs. circular tunnel). Hence, the results cannot clearly be related to one of these two factors. Hence, future research should focus on replicating the found effects when just one of factor (movement type or ID) is manipulated.

4.2 Motivation

In the present study, no correlation between participants’ motivation and the latency perception threshold could be found. However, it should be considered that all participants were strongly motivated to solve the task as good as possible. One explanation for the high motivation might be the sample itself. All participants were students who voluntarily took part in the experiment. It can be assumed that all participants had an above average tendency to strive for success, because they chose a demanding study with situations in which the individual performance is tested. Furthermore, it was clear that the individual latency perception threshold was tested in the present experiment which might led to a further self-selection of the sample. Hence, it is possible that the intrinsic motivation to solve the task as good as possible was very high for all participants and couldn’t be increased any further by the financial incentive [26].

Future research should focus on a more differentiated sample regarding the motivation to correctly identify the latencies to gather more insights to its role in this regard.

4.3 Visual Processing Speed

Results indicated that visual processing speed might be an important user characteristic with respect to latency perception. Nevertheless, it must be considered that the used tests examine the visual processing speed only indirectly.

Further tests should also directly measure the visual processing speed to replicate the found effects. A further research questions arises from the results that elderly people have more problems with latency perception [27] and that the processing speed decreases with higher age [28]. The questions arises, if the negative effects of age on latency perception are only caused by the reduced processing speed or if other variables (e.g., experience with computers) have an additional effect.

4.4 Conclusion

The present research found movement type and visual processing speed to be relevant influencing factors on latency perception in 2D-dragging tasks. The results should be considered in human-computer-design and provide implications to further research. Moreover, the identified latency perception thresholds in the present study were considerably under 100 ms (M = 49.63 ms; SD = 21.24), supporting the need for revised design guidelines regarding human-computer-interaction.