Effects of Overloading Command Capacity of Multiplexed FastTap

Menus on Spatial Learning in Tablets

Sayeem Md Abdullah

and Md. Sami Uddin

Department of Computer Science, University of Regina, Regina, Saskatchewan, Canada

Keywords: Command Selection, Spatial Memory Development, Menu Design, Tablets, Multi-Touch.

Abstract: Multiplexing can overload the command capacity of spatially stable tablet menus like FastTap by overlaying

multiple tabs of commands. While multiplexed menus can facilitate spatial learning and quick command

selections with a limited number of commands (20 items per tab), it is unclear whether multiplexed tablet

menus support spatial learning as the capacity of each tab increases. To that end, we conducted a controlled

study with four tab-based FastTap menus and investigated spatial learning in three sizes of tabs: Small (16

commands per tab), Medium (30), and Large (42). Results indicated that participants developed spatial

memory of commands in all conditions; however, the spatial memory development rate significantly slowed

down when the menu size grew. We discovered a reverse correlation between command capacity and spatial

memory development in multiplexed contexts, which could guide the design of future spatial memory

interfaces for tablets with increased command capacity.

1 INTRODUCTION

Multi-touch tablets require users to find and select

command locations, similar to any Graphical User

Interface (GUI). Spatial memory, which helps us

learn real-life locations, can be leveraged to design

interfaces that support rapid command selection

(Postma & De Haan, 1996). By displaying command

icons in fixed locations, spatial memory-based GUIs

can facilitate quick learning of command locations.

Research has shown that spatial memory techniques

can outperform popular command-selection tools

such as Ribbon menus, hierarchical menus, or

Marking Menus when using smaller command sets

(Gutwin & Cockburn, 2006; Mollashahi et al., 2018).

However, modern desktop GUIs like Microsoft

Office and Adobe Photoshop typically include

hundreds of commands in their menus, often unseen

on tablets. Moreover, displaying a large command set

on tablets can be difficult as they may slow down the

finding and selecting commands.

https://orcid.org/0009-0008-3324-5609

https://orcid.org/0000-0001-8354-2320

Figure 1: Command selections with FastTap. From left:

invoking a menu by pressing the menu button with the

thumb - novice selection; invoking the menu (thumb) and

incorrect command location (index); pressing the menu

(thumb) and correct command location (index) together

without invoking the entire menu - expert selection.

One reason for this difficulty could be the small

size of handheld devices, which may have made it

challenging to include many commands. Part of the

reason is that we know little about how increasing the

command capacity of tablet menus affects finding and

selecting commands. To address one aspect of this

problem, Gutwin et al. (2014) developed a rapid

Abdullah, S. M. and Uddin, M. S.

Effects of Overloading Command Capacity of Multiplexed FastTap Menus on Spatial Learning in Tablets.

DOI: 10.5220/0013319200003912

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 20th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2025) - Volume 1: GRAPP, HUCAPP

and IVAPP, pages 495-505

ISBN: 978-989-758-728-3; ISSN: 2184-4321

495

command selection technique called FastTap Menu

for tablets (see Figure 1).

They used a 5x4 grid to display 19 commands,

with one cell reserved for a menu button. Users could

access the commands by pressing the menu button

with their thumb and selecting an item with their

index finger while still pressing the menu button with

their thumb. Once learned, they could quickly execute

commands by combining menu invocation and

command selection into a thumb and index finger tap.

However, this menu had only 19 commands. Lately,

Abdullah & Uddin (2024a, 2024b) explored

increasing grid size to expand command capacity of

FastTap, and found that the larger grid could impede

spatial learning of commands.

In an effort to expand the command capacity in

tablets, Gaur et al. (2018) multiplexed the menu area

of FastTap. Their design used a tabbed layout with

four overlaying tabs, each containing a 20-item grid,

similar to the original FastTap design (Gutwin et al.,

2014). While results suggested that multiplexing

could potentially increase the command capacity of

FastTap Menus, it can create an additional level of

difficulty for learning and recalling command

locations. It is because each location in a multiplexed

menu contains multiple commands, overlaying on top

of each other, separated by tabs. Moreover, no prior

research explicitly explored the effect of overloading

command capacity on the development of spatial

memory. Without this knowledge, it would be

difficult to design better tablet interfaces that have

approximately 1.14 billion users worldwide (Mobile

App Statistics Everyone Should Know in 2023, n.d.).

Therefore, we conducted a study to test how

increasing command capacity affects spatial learning

when spatial memory was multiplexed with four tabs

within the FastTap Menus. During the study,

participants completed command selection tasks in

three different sizes of multi-tab FastTap menus:

Small (16 commands in one tab), Medium (30) and

Large (42). Small, medium, and large interfaces had

64, 120, and 168 commands, respectively, in total.

Key results from our study were:

 Users developed spatial memory of commands

in all interfaces; however, spatial learning

significantly impeded (increased completion

time and errors) when the menu grew;

 Although expert selections increased notably

over time, there was no difference in expert

selections across the three interfaces;

 Multiplexing spatial memory had fewer

interferences on learning, but interferences

increased with larger menus;

 Participants found interfaces more challenging

as menu items grew and preferred smaller.

Our research contributes several novel insights

into spatial learning in tablets. First, we conducted the

first study exploring how overloading command

capacity in multiplexed tablet menus impacts spatial

learning of commands. Second, we demonstrated that

users could develop spatial memory of commands

regardless of the command set size in multi-tab

tablets. However, we noticed a significant slowdown

in command finding and selecting performance as the

grid size increased. Lastly, our findings of the

negative correlation between menu multiplexing and

spatial learning can inform the design of future tablet

menus employing spatial memory multiplexing to

expand command capacity.

2 RELATED WORKS

2.1 Spatial Memory in GUIs

Spatial memory is one basic human ability that

supports learning and remembering information

about one’s surroundings and spatial orientation

(Kessels et al., 2002; Postma & De Haan, 1996).

Researchers have examined the development of

spatial memory (Postma & De Haan, 1996;

Thorndyke & Goldin, 1983). According to Siegel et

al.’s model (1975), spatial memory consists of three

types of knowledge: landmark, route, and survey.

Initially, people develop landmark knowledge

through visual search (Hasher & Zacks, 1979). As

they become more familiar, they acquire route

knowledge (Thorndyke & Hayes-Roth, 1982), which

then evolves into survey knowledge, enabling them to

recall information from memory.

Following real-life, the process of learning and

recalling information in GUIs usually involves three

stages: cognitive, associative, and autonomous, as

identified by Anderson (2000) and Fitts and Posner

(1967). During the cognitive stage, users acquaint

themselves with an interface and visually search for

commands. Then, in the associative stage, they

memorize commands and begin to retrieve them from

memory, although some visual searching may still

occur. Later in the autonomous stage, users can recall

commands without searching.

Researchers have utilized spatial memory in GUIs

to assist users in locating and selecting commands

quickly (Cockburn et al., 2014; Kurtenbach & Buxton,

1994; Zheng, Lewis, et al., 2018). Scarr et al.’s (2014)

CommandMaps for instance, present all commands in

spatially fixed locations in desktop GUIs, facilitating

HUCAPP 2025 - 9th International Conference on Human Computer Interaction Theory and Applications

496

faster retrieval, even for realistic tasks. Similarly,

Gutwin et al.’s (2006) ListMaps and Cockburn et al.’s

(2006) Space Filling Thumbnails have shown that

stable and flat menus can substantially improve item

revisitation performance in various contexts.

Spatial memory also enhances interactions on

multi-touch devices. Gutwin et al.’s (2014) FastTap

Menus use a stable flat grid menu on tablets, allowing

rapid command selection by combining thumb and

index finger actions. Similar methods relying on

spatial memory have been shown to enhance selection

speed on tablets (Gaur et al., 2018; Gutwin et al.,

2014; Uddin & Gutwin, 2016), smartwatches (Jannat

& Hasan, 2023), smartphones (Zhai & Kristensson,

2003; Zheng, Bi, et al., 2018), and tabletops (Uddin

et al., 2016). Moreover, studies indicate that spatial

memory provides benefits in large environments such

as VR (Gao et al., 2019), wall display (Jansen et al.,

2019), and even in touch tables (Joshi & Vogel, 2019).

However, these studies do not explore how varying

menu sizes affect the development of spatial memory.

2.2 Expanding Command Capacity in

Handheld Devices

Researchers explored various methods to increase the

command capacity of smartphones and tablets while

maintaining fast command selection speed. One

general approach was to use memory-based

mechanisms. For instance, Gutwin et al.’s (2014)

FastTap Menus employed a fixed grid to show 19

commands in tablets. An updated version of FastTap

expanded its capacity to 24 commands (Gutwin et al.,

2015). Abdullah & Uddin (2024a, 2024b) showed

that FastTap could hold up to 56 items by altering the

grid size and compromising spatial learning

performance. Another memory-based work by

Schramm et al. (2016) used a hidden bezel toolbar to

display 28 commands on tablets. Uddin et al. (2016),

however, relied on users’ proprioceptive knowledge

of hands to display 20 commands around the spread-

out fingers of a hand in tablets.

Using gesture-based methods was another way to

increase menu capacity on smaller devices. For

example, multi-touch marking menus (Lepinski et al.,

2010) can include about 64 commands. M3 gesture

menu (Zheng, Bi, et al., 2018) explored position-

specific gestures to overload command capacity.

Even augmented reality was investigated to virtually

display commands around smaller handheld devices,

which could accommodate a large set of commands

in a 46x27 grid (Hubenschmid et al., 2023).

Many people tried to increase command capacity

by using multiplexing menu space. For instance, Gaur

et al. (2018) significantly enhanced the command set

size of FastTap menus to 80 commands by

multiplexing the display space with four tabs, each

containing 20 commands. Uddin et al. (2016)

multiplexed the space between the index and thumb

to accommodate 80 commands with four tabs in

tablets. Even smartwatches can use multiplexing to

display up to 9 commands in a 3x3 grid (Lafreniere et

al., 2016). While multiplexing may overload the

capacity of FastTap, we still lack an understanding of

how it impacts the recall and learning of commands

across different sets. Thus, we intend to investigate

the influence of multiplexing on spatial learning.

3 STUDY INTERFACES

We aimed to explore how the increasing capacity of

commands influences the development of spatial

memory in menus with multiple tabs. Gaur et al.

(2018) extended the FastTap menu (Gutwin et al.,

2014) by including four tabs and demonstrated that

multiplexing could be an effective way to overload

the menu capacity of tablets.

We adapted Multi-tab FastTap menu to design

three prototype interfaces with three grid sizes: Small

(5x4), Medium (7x5), and Large (8x6), where we

used the bottom row for menu buttons (see Figure 2).

Following the original design, four cells of the bottom

row (starting from the left) were assigned to four tabs;

each tab could be accessed by tapping the respective

tab activation button. The rest of the bottom row was

blank. So, our three menus contained 16, 30, and 42

commands in each tab, respectively.

We used an 8.7-inch tablet (Samsung Galaxy Tab

A7 Lite) to implement our interfaces. Following Gaur

et al.’s (2018) Multi-tab FastTap and Parhi et al.’s

(2006) guide for touch target size, we initially

designed four grids, 5x4, 7x5, 9x6, and 11x7, with 64,

120, 192 and 280 items in total, respectively. During

a pilot study, people faced difficulty reaching the

targets displayed in the upper row, as our device was

larger than the original (8.7-inch instead of 7). Also,

people took significantly more time in the 11x7 grid.

So, we removed the 11x7 grid and reduced the Large

grid size to 8x6. Also, we reduced the grid height by

~1.7 inches from the top (that displayed target

questions) to ~7 inches, matching the original design.

So, in this study, we had three conditions (each

with four tabs), where each tab in the Small condition

had 16 items, resulting in a total of 64 items. The tabs

were arranged from left to right in the order of

Finance, Drinks, Clothing, and Applications.

Effects of Overloading Command Capacity of Multiplexed FastTap Menus on Spatial Learning in Tablets

497

Figure 2: Study interfaces with highlighted target locations. From top: Small (5x4), Medium (7x5) and Large (8x6).

Similarly, Medium had 30 items in a tab, and overall,

120 items, arranged in Transport, Household, Beauty,

and Produce tabs from left to right. The Large

condition included 42 items, resulting in 168 items in

total, organized in Academics, Flags, Fictional, and

Animals tabs, from left to right (see Figure 2).

4 STUDY METHOD

4.1 Tasks and Stimulus

In the study, participants selected a series of targets

that appeared at the top of the grid. Selection methods

were similar to FastTap (Gutwin et al., 2014), thumb

and index finger actions (see Figure 1). The default

setting concealed the commands, requiring a 200ms

press on the menu button with the thumb to reveal

them. Subsequently, users could select commands

with their index finger. Our interfaces supported

expert and novice selections (Gutwin et al., 2014). If

users know the menu, they can quickly select a

command without waiting 200ms for the whole menu

to display: an expert selection. Otherwise, users could

wait to see the entire menu and perform a visual

search for the command before selecting it: a novice

selection. Selection feedback was provided by

changing the selected cell’s background to green for

a correct selection; otherwise, it was red. Users could

move to the next target only after a successful

selection. They also had to lift their thumb from the

screen to progress to the next target question.

HUCAPP 2025 - 9th International Conference on Human Computer Interaction Theory and Applications

498

For the study, we selected four targets as stimuli

from each tab: one target had no overlap with any

other tab, two targets overlapped between two

different tabs, and one target overlapped in all four

tabs. Among the four targets, one was from the corner,

two were from the centre, and another one from the

side. This would allow us to determine any

interference due to multiplexing. So, in total, sixteen

targets were used as stimuli in a condition.

During the study, trials were repeated over

eighteen blocks, where after every fifth block, we

kept a blind block, where the menu elements

remained hidden even after pressing the menu button.

The three blind blocks required participants to rely on

memory to remember target positions, encouraging

people to become familiar with the menu. In total,

there were fifteen regular and three blind blocks.

4.2 Procedure and Study Design

We conducted a within-subject study with three

conditions: Small (5x4), Medium (7x5), and Large

(8x6) grids. Each condition had four tabs, with four

different command sets (Figure 2). The study

consisted of 18 blocks of trials, where one block

included sixteen targets (order randomized). The

order of the conditions was balanced using a Latin

Square design (Grant, 1948). Participants were

instructed to complete tasks with speed and accuracy.

Participants completed a practice session with a

4x3 grid to familiarize themselves with the interface

and operations. The practice session included three

tabs: Zodiac, Folder, and Music, arranged from left to

right. Participants completed 18 blocks of trials,

including three blind blocks, with three targets. After

completing each condition, participants completed a

NASA-TLX questionnaire, and after completing all

four conditions, they provided their preferences.

4.3 Participants and Apparatus

Fourteen individuals (twelve Cis men, two Cis

women, one left-handed, ages 19-33, mean age 24.21)

were recruited from a local university to participate in

the study. All were familiar with touch devices, with

an average of 9.29 years of experience. Furthermore,

the average screen time of participants was 5.46 hours

per day. The study session lasted ~60 minutes, and

each participant received a $10 gift card as an

honorarium.

All experiments were conducted on a Samsung

Galaxy Tab A7 Lite 8.7-inch multi-touch display. The

application was developed for Android using Java.

5 RESULTS

Due to technical and operational issues, a few trials

needed excessive time to complete. So, we excluded

168 out of 10080 (1.67%) from the regular blocks and

42 out of 2016 (2.08%) from the blind blocks as

outliers (3 s.d. away from the means of respective

blocks). Shapiro-Wilk tests indicated that our samples

followed an approximately normal distribution,

confirmed by Q-Q plots. Therefore, we proceeded

with ANOVA for our analyses. We report the effect

size for significant RM-ANOVA results as general

eta-squared: η

(considering .01 small, .06 medium,

and >.14 large (Cohen, 1973)), and Bonferroni

correction was performed for post-hoc pairwise t-

tests. We performed Greenhouse-Geisser adjustments

on the results of our study (resulting in fractional

degrees of freedom), where ANOVA’s sphericity

assumption was violated (Mauchley’s test).

5.1 Trial Completion Time

The trial completion time was calculated from when

a target question appeared on the screen to when a

user successfully selected the required target. Mean

completion times are summarized in Figure 3 for

regular and blind blocks (B6, B12, and B18). For

regular blocks, ANOVA showed a significant main

effect of condition (F

2,26

=26.87, p<0.001, η² =0.22).

Figure 3: Trial Completion Time (±s.e.) by Condition and

Block.

Small condition (mean 2463.43ms, s.d.1335.25)

outperformed Medium (3039.46ms, s.d. 1992.43),

and Large (3654.92ms, s.d. 2818.56). People learned

command locations quickly; hence, completion time

dramatically decreased across blocks (F

14,182

=46.96,

p<0.001, η² =0.61). We also found a condition ×

block interaction (F

28,364

=8.36, p<0.001, η² =0.22).

A similar pattern was found in blind blocks, where

ANOVA found a significant main effect of condition

2,26

= 6.80, p<0.001, η²=0.27), block (F

2,26

=35.85,

Effects of Overloading Command Capacity of Multiplexed FastTap Menus on Spatial Learning in Tablets

499

p<0.001, η²=0.23), and condition × block (F

4,52

=1.73,

p=0.16) with Small (mean 2114.23ms, s.d. 1137.32),

Medium (2492.72ms, s.d. 1304.71), and Large

(2840.03ms, s.d. 1676.46). Follow-up tests also

showed significant differences between all the

conditions for both regular and blind blocks (all

p<0.001), indicating consistent results throughout.

Additional analysis on regular blocks by condition

and position revealed that there were differences in

mean completion time by positions: corner (mean:

2800.39ms, s.d. 2574.22), centre (3074.88ms, s.d.

2387.67) and side (3043.58ms, s.d. 2656.09).

ANOVA found an effect of position (F

2, 26

=16.04,

p<0.001, η²=0.09), condition (F

2,26

=23.24, p<0.001,

η²=0.56) and condition × position interaction (F

4,52

=11.40, p<0.001, η²=0.15) on completion time in

regular blocks. However, follow-up tests show

significant differences between corner:centre and

corner:sides only across three conditions (p<0.007).

5.2 Expert Selections

We recorded expert selections when a user selected a

target before it appeared on the grid (less than 200ms),

irrespective of accuracy. Figure 4 illustrates the

results of expert selections in regular blocks.

Figure 4: Expert Selection (±s.e.) by Condition and Block.

Although expert selections significantly increased

over time, block (F

2.29, 29.81

=11.68, p<0.001, η²=0.29),

with over 60% expert selections in later blocks,

ANOVA showed no effect of condition (F

2,26

<0.001,

p=0.99) and condition × block interaction

28,364

=0.64, p=0.02). Mean expert selection rates

were identical across conditions: Small (mean: 0.51,

s.d. 0.62), Medium (mean: 0.51, s.d. 0.85) and Large

(mean: 0.51, s.d. 0.80).

Additional analysis shows that out of 9912

successful selections made during regular blocks,

3698 (37.31%) were experts. Among the 1903

erroneous selections, 1439 (75.62%) were erroneous

expert selections. When categorized according to the

menu size (Small, Medium, and Large), there were

1409, 1180, and 1109 successful expert selections,

and 301, 536, and 602 unsuccessful expert selections,

respectively. So, the results indicate that larger

multiplexed menus could impede performance.

Further analysis by condition and position showed

no significant effect of any measures: condition (F

=0.02, p=0.98), position (F

2, 26

=2.10, p=0.14), and

condition × position (F

4,52

=0.64, p=0.63), with

corner (mean: 0.48, s.d. 0.70), centre (0.54, s.d. 0.86)

and side (0.51, s.d. 0.88).

5.3 Errors in Selection

The errors in selection were calculated by the number

of incorrect selections before the correct one, which

are summarized in Figure 5.

Figure 5: Errors (±s.e.) by Condition and Block.

ANOVA showed a significant effect of condition

on errors (F

2,26

=4.76, p=0.01, η²=0.06), with Small

being the most accurate (mean errors 0.13, s.d. 0.42),

Medium (0.21, s.d. 0.69), and Large having the most

errors (0.23, s.d. 0.63). As seen in Figure. 5, errors

increased slightly in the later blocks, yielding a

significant effect of block (F

14,182

=4.31, p<0.001,

η²=0.09). A probable reason could be the reliance on

weakly developed spatial memory (due to larger

menu sizes) to recall target positions. We found no

condition × block interaction (F

28, 364

=1.04, p=0.42).

Post-hoc pairwise tests showed significant

differences between all conditions (all p<0.001)

except Medium and Large (p=0.12).

Similar results were found in blind blocks

providing significant effects of condition (F

1.22,

15.90

=4.88, p=0.04, η²=0.18), block (F

1.18,15.31

=10.47,

p=0.004, η²=0.12), and condition × block interaction

4,52

=2.85, p=0.03, η²=0.05). Mean errors being the

lowest in Small (mean 0.19, s.d. 0.55), Medium (0.49,

s.d. 1.22), and the highest in Large (0.93, s.d. 2.40),

all pairs being notably different (all p<0.002).

HUCAPP 2025 - 9th International Conference on Human Computer Interaction Theory and Applications

500

Table 1: Error analyses by condition and overlapped tabs.

Small Medium Large

Overall No

Overlap

2 Tabs

Overlap

4 Tabs

Overlap

Overall No

Overlap

2 Tabs

Overlap

4 Tabs

Overlap

Overall No

Overlap

2 Tabs

Overlap

4 Tabs

Overlap

Correct

(correct tab &

position)

2959

(87.41%)

789

(95.98%)

1443

(85.23%)

727

(83.56%)

2865

(80.42%)

723

(83.39%)

1452

(80.76%)

690

(76.90%)

2771

(78.06%)

729

(84.47%)

1362

(76.27%)

680

(75.46%)

Off-by-1 error

(correct tab, 1

pos. off)

315

(9.31%)

(3.04%)

179

(10.57%)

111

(12.76%)

408

(11.44%)

(9.11%)

197

(10.96%)

132

(14.65%)

475

(13.38%)

(7.19%)

265

(14.83%)

148

(16.43%)

Other error

(correct tab, >1

pos. off)

(1.27%)

(0.61%)

(1.54%)

(1.38%)

125

(3.51%)

(2.77%)

(3.73%)

(3.77%)

148

(4.17%)

(3.13%)

(4.14%)

(5.22%)

Tab error

(incorrect tab,

correct pos.)

(1.33%)

(0.12%)

(1.77%)

(1.61%)

(2.16%)

(2.42%)

(2.00%)

(2.23%)

(1.97%)

(2.20%)

(1.96%)

(1.78%)

Tab + off-by-1

(incorrect tab,

1 pos. off)

(0.50%)

(0%)

(0.71%)

(0.57%)

(1.77%)

(1.38%)

(1.83%)

(2.00%)

(1.32%)

(1.27%)

(1.68%)

(0.67%)

Tab + other

(incorrect tab,

>1 pos. off)

(0.18%)

(0.24%)

(0.18%)

(0.11%)

(0.70%)

(0.92%)

(0.72%)

(0.45%)

(1.10%)

(1.74%)

(1.12%)

(0.44%)

We also analyzed the regular blocks based on their

condition and position. ANOVA showed no

significant effect of position (F

2,26

=2.10, p=0.14),

condition (F

2,26

=0.02, p=0.98) and condition ×

position interaction (F

4,52

=0.64, p=0.63). However,

the corner had the most accuracy (mean errors: 1.41,

s.d. 0.80), followed by the centre (1.52, s.d. 1.00), and

the side had the least accuracy (1.99, s.d. 2.10).

Furthermore, we analyzed the errors in the regular

blocks based on different conditions and overlapped

tabs, excluding blocks B6, B12, and B18 (see Table

1). Our analysis suggests that participants made more

errors in the Large condition than others.

We observed that for tab-related errors, where

incorrect menus/tabs were selected but grid-locations

of targets were correct or missed by 1 cell, overall,

multiplexed spatial menus yielded relatively low

errors: less than 4% across three conditions –

indicating generally fewer tab-related interferences.

However, these interferences increased when the

menu size grew. We further noticed that most errors

in the study were off-by-1 (correct tab, 1 pos. off).

This means that participants selected the correct tab,

but selections were one position off. The least

common errors in this category were from Tab +

other (incorrect tab, >1 pos. off), which is when the

participant selected the incorrect tab, and selections

were more than one position off. Again, results

indicate that these errors also increased when menu

sizes increased; Large condition produced the highest

number of errors.

5.4 Subjective Responses

We compared the raw NASA-TLX responses for all

the conditions (see Table 2).

Table 2: Error Mean (s.d.) effort scores (1-10 scale, low to

high; performance reversed, failure to perfect).

Small Medium Large

𝓧

𝒓

𝟐

Mental

5.86

(2.41)

7.36

(2.21)

8.29

(1.59)

26.46 0.01

Physical

4.79

(2.55)

5.29

(2.76)

6.64

(2.24)

30.79 0.004

Temporal

5.57

(2.68)

5.07

(2.59)

7.07

(2.27)

28.30 0.01

Performance

8.43

(1.65)

7.93

(1.69)

7.50

(1.79)

25.29 0.02

Effort

5.93

(1.90)

6.79

(2.19)

7.79

(1.93)

27.87 0.01

Frustration

4.14

(2.80)

5.29

(2.73)

6.14

(3.08)

29.60 0.01

Results indicated a significant difference between

all the conditions, with the Large condition having a

higher overall workload score than others. Post-hoc

tests revealed a significant difference in mental

demand between Small and Large conditions

(p=0.01). However, no significant differences were

found in physical, temporal, performance, effort, and

frustration across all conditions.

Effects of Overloading Command Capacity of Multiplexed FastTap Menus on Spatial Learning in Tablets

501

Table 3: Participant preferences among conditions.

Small Medium Large

Speed 8 5 1

Accuracy 6 8 0

Memorization 9 4 1

Expert Mode 5 5 4

Comfort 10 3 1

Overall 3 10 1

Participant preferences are reported in Table 3.

Regarding speed, memorization and comfort, at least

57% of participants preferred the Small condition,

and out of the two larger conditions, Medium was the

second-best preferred menu. In terms of accuracy,

interestingly, 58% of users preferred Medium, and the

rest chose Small. For expert selection use, 4 people

preferred the Large condition. Finally, in the case of

overall satisfaction, interestingly, the majority, 71%,

preferred Medium over Small.

6 DISCUSSIONS

6.1 Explanations for Results

6.1.1 All Overloaded Menus Enabled Spatial

Memory Development

Our analyses indicated that the mean trial completion

time decreased across the blocks, as depicted in

Figure 3, despite differences in command set sizes. It

suggests that people developed spatial memory in all

conditions, which can be explained by the three stages

of spatial learning (Fitts & Posner, 1967).

Our participants were unfamiliar with the three

multiplexed FastTap menus, causing them to spend

more time searching for the targets visually. At this

time, they were actively engaged in the cognitive

stage of spatial learning (Fitts & Posner, 1967). Their

effortful menu interactions while searching for the

targets could also contribute to the spatial learning of

commands as a by-product (Cockburn, Kristensson,

et al., 2007; Darken & Sibert, 1996). During the study,

our participants might have transitioned to the

associative stage of spatial learning after a short

interaction with the menus (Fitts & Posner, 1967). We

also observed a dramatic reduction in completion

time during the early stages of our study (around 3-5

blocks, as shown in Figure 3). Participants began to

utilize more expert selections at these stages (Figure

4). They skipped waiting (200ms) for command icons

to appear and instead relied on their memory to recall

target locations in all menus. This suggests that

participants in our study developed spatial memory

for commands across all overloaded menus. Our

findings also indicated that participants could utilize

their spatial knowledge more effectively and

efficiently in smaller menus than larger menus with

more items, as demonstrated by completion time and

error rate analyses.

6.1.2 Why Did Selection Performance

Increase as Time Progressed?

We observed a significant improvement in users’

selection performance over time, as evidenced by

reduced trial completion time and increased expert

selections. We believe FastTap’s selection

mechanism contributed to this performance growth.

The target selection process involved three steps.

First, to invoke the hidden menu, participants placed

their thumb on the Menu buttons at the bottom row.

Second, when the menu was displayed on the screen,

people could visually search for a desired item.

Finally, people could use their index finger to select

the target while keeping their thumb on the menu.

Novice users must follow these three steps to select

an item if unfamiliar with the target locations.

However, with practice and over time, users could

learn the locations of commands and quickly recall

them from memory. It allowed them to merge the

initial and final steps, skipping the visual search and

promptly selecting the target commands (see Figure

1). It is worth noting that novice users in our study

also experienced a 200ms delay between the menu

button press and the commands display. Expert users,

however, could skip this delay by merging menu

activation and target selection into a chunked thumb-

index action. Though menu sizes varied in our study

and involved multiplexing the menu space, this

pattern was evident in all three sizes of menus.

6.1.3 Does Multiplexing Large Menus

Impede Spatial Learning?

While all menus in our study facilitated spatial

memory development regardless of their size, we

observed a significant reduction in spatial learning as

the menu size increased. We identify two potential

reasons behind these findings.

First, the choice-reaction time of Hick-Hayman

Law (Cockburn, Gutwin, et al., 2007; Hick, 1952) and

the target size of Fitts’ Law (1954) can explain the

relatively slower target selection speed in larger

interfaces. Since the number of commands increased

HUCAPP 2025 - 9th International Conference on Human Computer Interaction Theory and Applications

502

in larger menus, from 16 to 30 and 42 in one grid, the

available choices for users to select from increased

(and the size of targets decreased) substantially. In

addition, our menus had four tabs, providing

additional choices at the tab level. Due to

multiplexing, each grid cell included four overlapping

commands. As such, people may have spent

substantially more time searching for the target

locations and correct tab. Additionally, the smaller

target sizes in larger menus may have made selection

difficult. FastTap enables rapid selections based on

spatial memory, but the presence of menu and

command choices from multiple large command sets

may have hindered its quick development.

Second, people typically rely on landmarks

present in GUIs (Uddin & Gutwin, 2021) to learn

command locations. In our study, grid lines, corners

and bezels of the tablet could work as landmarks and

support spatial learning, at least in smaller conditions.

However, when the number of commands increased,

and menus were multiplexed, existing landmarks may

have weakened, as one landmark acted as a reference

point for multiple and overlapping items. Our

position-specific analyses suggested that people

mostly struggled and made more errors with targets

located in the centre and side, particularly in larger

menus. Using additional landmarks (Uddin, 2022)

could improve spatial learning in larger-size menus

and could be a future avenue of research.

6.2 Design Implications and

Generalizing Results

Our results suggest that participants had difficulty

developing spatial knowledge of commands as the

menu size increased, resulting in slower selection

speed and more errors in larger-multiplexed

interfaces. Therefore, designers can rely on

multiplexing menu space to increase the command

capacity of tablet menus without compromising

performance, as long as the grid size remains small.

They need to be cautious when overloading the menu

capacity. However, in such situations, our results

suggest that designers can trade among three regions

of a screen: sides, corners, and centre. Although

selection performance decreased when menu size

grew, we found that targets from the corner region

required significantly less time than the side and

centre regions. In addition, our results raise the

question of whether multiplexing alone is adequate to

increase command capacity in menus, particularly

with larger command sets. While our results

demonstrated the value of multiplexing in enabling

spatial learning, as discussed above, the value could

be further enhanced by adding spatial learning

support, like landmarks, in larger menus, a route we

aim to explore in future.

6.3 Limitations and Future Work

Our research has limitations that we plan to tackle in

future. First, following prior research on FastTap, we

developed prototype interfaces on an 8.7-inch tablet,

maintaining a 7-inch grid size, which yielded

promising initial results. Moving forward, we plan to

explore multiple avenues. The goals include

developing real-world applications, conducting long-

term studies, and implementing Multi-tab FastTap

Menus across various screen sizes for performance

assessment. We plan to integrate this technology into

new foldable devices like the Galaxy Z Fold and

Google Pixel Fold to improve their capabilities.

Second, to minimize interference among tabs and

ensure clarity in distinguishing each tab, we utilized

twelve distinct sets of icons (Figure 2). This design

aimed to prevent confusion and support learning by

providing unique visual references for each tab. The

familiarity of these icon sets among participants may

have influenced selection performance, as the study

used icon names as target stimuli. For instance,

participants might find it easier to recognize Animal

icons than Fictional icons, potentially impacting

target icon search performance. Future research could

explore how familiarity influences cognitive

strategies, spatial learning, and decision-making

processes in similar contexts.

7 CONCLUSIONS

We studied the effect of overloading menu capacity

on spatial memory in tablets using multiplexed

FastTap Menus with four tabs in three different grid

sizes: Small, Medium, and Large, containing 64, 120,

and 168 items, respectively. While spatial learning

dramatically slowed in larger menus, we found that

people developed spatial memory in all multiplexed

menus. As the grid size grew, both selection time and

errors rose, but participants still effectively used the

expert selection mode. Our research provides new

empirical evidence showing an inverse relationship

between menu capacity and spatial learning of

commands. These findings can form the basis for

developing future tablet interfaces with extensive

command sets by multiplexing spatial memory.

Effects of Overloading Command Capacity of Multiplexed FastTap Menus on Spatial Learning in Tablets

503

ACKNOWLEDGEMENTS

The work was supported by the Natural Sciences and

Engineering Research Council of Canada (RGPIN-

2023-04393).

REFERENCES

Abdullah, S. M., & Uddin, Md. S. (2024a). Revisiting

FastTap: Effects of Increasing Command Capacity of

Spatial Memory Menus in Tablets. 2024 IEEE 15th

Annual Ubiquitous Computing, Electronics &; Mobile

Communication Conference (UEMCON), 0683–0689.

https://doi.org/10.1109/UEMCON62879.2024.107546

Abdullah, S. Md., & Uddin, Md. S. (2024b). Effects of

Increasing Command Capacity of Spatial Memory

Menus in Tablets. Companion Proceedings of the 2024

Conference on Interactive Surfaces and Spaces, 69–72.

https://doi.org/10.1145/3696762.3698055

Anderson, J. R. (2000). Learning and memory: An

integrated approach, 2nd ed. In Learning and memory:

An integrated approach, 2nd ed. John Wiley & Sons

Inc.

Cockburn, A., Gutwin, C., & Alexander, J. (2006). Faster

document navigation with space-filling thumbnails.

Conference on Human Factors in Computing Systems -

Proceedings, 1, 1–10. https://doi.org/10.1145/11247

72.1124774

Cockburn, A., Gutwin, C., & Greenberg, S. (2007). A

predictive model of menu performance. Proceedings of

the SIGCHI Conference on Human Factors in

Computing Systems - CHI ’07, 627–636.

https://doi.org/10.1145/1240624.1240723

Cockburn, A., Gutwin, C., Scarr, J., & Malacria, S. (2014).

Supporting Novice to Expert Transitions in User

Interfaces. ACM Computing Surveys, 47(2), 1–36.

https://doi.org/10.1145/2659796

Cockburn, A., Kristensson, P. O., Alexander, J., & Zhai, S.

(2007). Hard lessons: Effort-inducing interfaces benefit

spatial learning. Conference on Human Factors in

Computing Systems - Proceedings, 1571–1580.

https://doi.org/10.1145/1240624.1240863

Cohen, J. (1973). Eta-squared and partial eta-squared in

communication science. Human Communication

Research, 28(56), 473–490.

Darken, R. P., & Sibert, J. L. (1996). Wayfinding strategies

and behaviors in large virtual worlds. Conference on

Human Factors in Computing Systems - Proceedings,

142–149. https://doi.org/10.1145/238386.238459

Fitts, P. M. (1954). The information capacity of the human

motor system in controlling the amplitude of

movement. Journal of Experimental Psychology, 47(6),

381–391. https://doi.org/10.1037/H0055392

Fitts, P. M., & Posner, M. I. (1967). Human performance.

In Human performance. Brooks/Cole.

Gao, B., Kim, B., Kim, J.-I., & Kim, H. (2019).

Amphitheater Layout with Egocentric Distance-Based

Item Sizing and Landmarks for Browsing in Virtual

Reality. International Journal of Human–Computer

Interaction, 35(10), 831–845. https://doi.org/10.1080/

10447318.2018.1498654

Gaur, V., Uddin, Md. S., & Gutwin, C. (2018).

Multiplexing spatial memory: increasing the capacity

of FastTap menus with multiple tabs. Proceedings of

the 20th International Conference on Human-

Computer Interaction with Mobile Devices and

Services - MobileHCI ’18, 1–13. https://doi.org/10.11

45/3229434.3229482

Grant, D. A. (1948). The latin square principle in the design

and analysis of psychological experiments.

Psychological Bulletin, 45(5), 427–442.

https://doi.org/10.1037/H0053912

Gutwin, C., & Cockburn, A. (2006). Improving list

revisitation with ListMaps. Proceedings of the

Workshop on Advanced Visual Interfaces, 2006, 396–

403. https://doi.org/10.1145/1133265.1133347

Gutwin, C., Cockburn, A., & Lafreniere, B. (2015). Testing

the rehearsal hypothesis with two FastTap interfaces.

Proceedings of the 41st Graphics Interface Conference

- GI ’15, 223–231.

Gutwin, C., Cockburn, A., Scarr, J., Malacria, S., & Olson,

S. C. (2014). Faster command selection on tablets with

FastTap. Proceedings of the ACM Conference on

Human Factors in Computing Systems - CHI ’14,

2617–2626. https://doi.org/10.1145/2556288.2557136

Hasher, L., & Zacks, R. T. (1979). Automatic and effortful

processes in memory. Journal of Experimental

Psychology: General, 108(3), 356–388. https://doi.org/

10.1037/0096-3445.108.3.356

Hick, W. E. (1952). On the rate of gain of information.

Quarterly Journal of Experimental Psychology, 4(1),

11–26. https://doi.org/10.1080/17470215208416600

Hubenschmid, S., Zagermann, J., Leicht, D., Reiterer, H.,

& Feuchtner, T. (2023). ARound the Smartphone:

Investigating the Effects of Virtually-Extended Display

Size on Spatial Memory. Proceedings of the 2023 CHI

Conference on Human Factors in Computing Systems,

1–15. https://doi.org/10.1145/3544548.3581438

Jannat, M. E., & Hasan, K. (2023). Exploring the Effects of

Virtually-Augmented Display Sizes on Users’ Spatial

Memory in Smartwatches. Proceedings - 2023 IEEE

International Symposium on Mixed and Augmented

Reality, ISMAR 2023, 553–562. https://doi.org/10.11

09/ISMAR59233.2023.00070

Jansen, Y., Schjerlund, J., & Hornbæk, K. (2019). Effects

of Locomotion and Visual Overview on Spatial

Memory when Interacting with Wall Displays.

Proceedings of the 2019 CHI Conference on Human

Factors in Computing Systems - CHI ’19, 1–12.

Joshi, N., & Vogel, D. (2019). An Evaluation of Touch

Input at the Edge of a Table. Proceedings of the 2019

CHI Conference on Human Factors in Computing

Systems - CHI ’19, 1–12.

Kessels, R. P. C., Jaap Kappelle, L., de Haan, E. H. F., &

Postma, A. (2002). Lateralization of spatial-memory

HUCAPP 2025 - 9th International Conference on Human Computer Interaction Theory and Applications

504

processes: evidence on spatial span, maze learning, and

memory for object locations. Neuropsychologia, 40(8),

1465–1473.

Kurtenbach, G., & Buxton, W. (1994). User learning and

performance with marking menus. Proceedings of the

SIGCHI Conference on Human Factors in Computing

Systems Celebrating Interdependence - CHI ’94, 258–

264. https://doi.org/10.1145/191666.191759

Lafreniere, B., Gutwin, C., Cockburn, A., & Grossman, T.

(2016). Faster Command Selection on Touchscreen

Watches. Proceedings of the ACM Conference on

Human Factors in Computing Systems - CHI ’16,

4663–4674. https://doi.org/10.1145/2858036.2858166

Lepinski, G. J., Grossman, T., & Fitzmaurice, G. (2010).

The design and evaluation of multitouch marking

menus. Proceedings of the ACM Conference on Human

Factors in Computing Systems - CHI ’10, 2233–2242.

https://doi.org/10.1145/1753326.1753663

Mobile App Statistics Everyone Should Know in 2023.

(n.d.). Retrieved April 4, 2024, from

https://techreport.com/statistics/app-statistics/

Mollashahi, E. S., Sami Uddin, M., & Gutwin, C. (2018).

Improving revisitation in long documents with two-

level artificial-landmark scrollbars. Proceedings of the

Workshop on Advanced Visual Interfaces AVI.

https://doi.org/10.1145/3206505.3206554

Parhi, P., Karlson, A. K., & Bederson, B. B. (2006). Target

size study for one-handed thumb use on small

touchscreen devices. ACM International Conference

Proceeding Series, 159, 203–210. https://doi.org/

10.1145/1152215.1152260

Postma, A., & De Haan, E. H. F. (1996). What was where?

Memory for object locations. The Quarterly Journal of

Experimental Psychology. A, Human Experimental

Psychology, 49(1), 178–199. https://doi.org/10.1080/

713755605

Scarr, J., Cockburn, A., Gutwin, C., Bunt, A.,

Cechanowicz, J. E., Scarr, J., Cockburn, A., Gutwin, C.,

Bunt, A., & Cechanowicz, J. E. (2014). The usability of

CommandMaps in realistic tasks. Proceedings of the

32nd Annual ACM Conference on Human Factors in

Computing Systems - CHI ’14, 2241–2250.

Schramm, K., Gutwin, C., & Cockburn, A. (2016).

Supporting Transitions to Expertise in Hidden

Toolbars. Proceedings of the SIGCHI Conference on

Human Factors in Computing Systems - CHI ’16,

4687–4698. https://doi.org/10.1145/2858036.2858412

Siegel, A. W., & White, S. H. (1975). The Development of

Spatial Representations of Large-Scale Environments.

Advances in Child Development and Behavior, 10, 9–

55. https://doi.org/10.1016/S0065-2407(08)60007-5

Thorndyke, P. W., & Goldin, S. E. (1983). Spatial Learning

and Reasoning Skill. In Spatial Orientation (pp. 195–

217). Springer US. https://doi.org/10.1007/978-1-

4615-9325-6_9

Thorndyke, P. W., & Hayes-Roth, B. (1982). Differences in

spatial knowledge acquired from maps and navigation.

Cognitive Psychology, 14

(4), 560–589. https://doi.org/

10.1016/0010-0285(82)90019-6

Uddin, M. S. (2022). Use of Landmarks to Improve Spatial

Learning and Revisitation in Computer Interfaces.

University of Saskatchewan.

Uddin, Md. S., & Gutwin, C. (2016). Rapid Command

Selection on Multi-Touch Tablets with Single-Handed

HandMark Menus. Proceedings of the 2016 ACM

International Conference on Interactive Surfaces and

Spaces, 205–214. https://doi.org/10.1145/2992154.29

92172

Uddin, Md. S., & Gutwin, C. (2021). The Image of the

Interface: How People Use Landmarks to Develop

Spatial Memory of Commands in Graphical Interfaces.

Proceedings of the 2021 CHI Conference on Human

Factors in Computing Systems, 1–17. https://doi.org/

10.1145/3411764.3445050

Uddin, Md. S., Gutwin, C., & Lafreniere, B. (2016).

HandMark Menus: Rapid Command Selection and

Large Command Sets on Multi-Touch Displays.

Proceedings of the 2016 CHI Conference on Human

Factors in Computing Systems, 5836–5848.

https://doi.org/10.1145/2858036.2858211

Zhai, S., & Kristensson, P.-O. (2003). Shorthand writing on

stylus keyboard. Proceedings of the Conference on

Human Factors in Computing Systems - CHI ’03, 97–

104. https://doi.org/10.1145/642611.642630

Zheng, J., Bi, X., Li, K., Li, Y., & Zhai, S. (2018). M3

Gesture Menu: Design and Experimental Analyses of

Marking Menus for Touchscreen Mobile Interaction.

Proceedings of the Conference on Human Factors in

Computing Systems - CHI ’18, 1–14. https://doi.org/

10.1145/3173574.3173823

Zheng, J., Lewis, B., Avery, J., & Vogel, D. (2018).

FingerArc and FingerChord: Supporting Novice to

Expert Transitions with Guided Finger-Aware

Shortcuts. The 31st Annual ACM Symposium on User

Interface Software and Technology - UIST ’18, 347–

363. https://doi.org/10.1145/3242587.3242589

Effects of Overloading Command Capacity of Multiplexed FastTap Menus on Spatial Learning in Tablets

505