The Cognitive Load Limits of Multiple Safety Signs

Abstract

Current construction safety practices heavily rely on the use of multiple safety signs to mitigate potential risks. This study investigates the cognitive load imposed on construction workers by various design characteristics of safety signs, including text-only, pictogram with text, and pictogram-only signs. A comprehensive cognitive assessment of 513 construction workers was conducted to measure recall rates and visual attention. The results revealed that text-only signs led to lower cognitive load compared with pictogram-based signs, with no significant split-attention effect observed when text and pictograms were combined. The average recall rate across all sign types was 3.32 items, with over 95% of participants recalling six or fewer items. Additionally, recall rates for pictogram-based signs decreased significantly when more than nine items were displayed. Visual attention analysis indicated that while color had no significant impact, the order of placement did. These findings highlight the importance of limiting the number of safety signs and carefully designing them to optimize recall and reduce cognitive load. Future research should explore additional factors such as sex, complexity of work, and environmental conditions to develop a more comprehensive understanding of cognitive load in construction settings.

1. Introduction

In construction sites, workers are often engaged in various hazardous tasks under complex work environments. To mitigate the potential risks, it is a standard practice to post multiple safety signs at work sites. However, given that human information processing capacity is finite, displaying an excessive amount of information may lead to cognitive overload. As a result, the effectiveness of conveying information about potential risks at the site may be negatively impacted, resulting in unsafe behavior.

Cognitive limitations have been identified as one of the factors contributing to unsafe behaviors in workers. In their research, Nwagbala and Park (2023) [1] investigated the influence of the foreman’s behavior on the cognitive processes of workers and demonstrated that any negative behavior on the part of the foreman can lead to cognitive failure and unsafe behavior by the worker. Similarly, Chen et al. (2016) [2] found that when workers in complex construction tasks are exposed to unforeseen hazards, cognitive limitations can impact their risk perception and lead to unsafe behaviors, which significantly contribute to accidents. In addition, Kim and Kim (1990) [3] emphasized that unsafe behavior is a result of human error, which arises from the limitations of human cognition and attention.

To combat unsafe behaviors in the workplace, the use of safety signs has been shown to be an effective strategy by drawing workers’ attention and heightening their risk perception [4]. Safety signs play a critical role in accident prevention by providing visual information about residual risk factors at work sites, drawing workers’ attention to potential risks, and assisting them in making informed safety decisions [5]. Additionally, safety signs are essential in enforcing compliance by effectively conveying hazard and risk information and lawful requirements through pictograms and text to workers and site visitors. They also play a crucial role in displaying emergency response procedures. Safety signs are arguably one of the most effective tools in risk communication, contributing to safe behavior decisions by providing workers with information on a task’s inherent hazards, influencing behavior, and serving as reminders [6].

Due to the hazardous nature of construction sites and in compliance with the safety signposting regulations of the Occupational Safety and Health Act, multiple safety signs are posted at several locations in work sites, construction equipment, and handling facilities to provide information about the hazards and dangers associated with the work (Figure 1).

However, excessive posting of safety signs can result in cognitive overload due to the limitations of human cognitive processing capabilities, which can hinder the effective comprehension of risk information and the induction of safe behavior.

To examine the cognitive load limit of workers in relation to safety sign postings on construction sites, the present study employed a cognitive psychology approach. This approach investigated the impact of multiple safety sign postings and the influence of design factors of safety signs (pictogram, text, location, and color) on the recall rates and visual attention of construction workers, with an aim to present an effective method of information transmission.

The subsequent sections of this study will look at the supporting literature, the methodology, and the result of the approach.

2. Literature Review

2.1. Human Information Processing

Humans receive external stimuli and information by interacting with their surroundings, and the received information is processed in the human cognitive memory system. About 60% of the information is received through visual stimuli and 20% through auditory stimuli, making sight an extremely important sense and the most frequently used one [7]. Atkinson and Shiffrin’s (1968) multi-store memory model asserts that the memory system consists of several storage units including sensory memory, short-term memory, and long-term memory. Human cognitive memory processing involves a sequence of stages arranged in the flow of time, with information being processed through rehearsal, as established in the modal model [8].

The cognitive phenomenon of human memory processing encompasses an intricate sequence involving encoding, storing, retaining, and retrieving information received through sensory organs. In human information processing theory, various memory classifications are identified, encompassing sensory memory as a distinct category. The sensory memory function is to briefly retain stimuli originating from sensory organs. This process facilitates the entry of information into human consciousness, where it is transiently stored through the mechanism of short-term memory. Additionally, working memory retains and processes information that has passed through sensory and short-term memory for a short period, and long-term memory has an unlimited capacity and stores information for a lifetime, available for retrieval at any time [9].

Working memory refers to a processing system in which limited information is consciously perceived for a short duration. Some view working memory and short-term memory as the same, while others see working memory as an intermediate stage developing into short-term memory. It is a memory used in the process of carefully handling specific tasks, distinguished from short-term memory by its active memory system for handling, storing, and processing [10].

When understanding sentences, the meanings of words must be retrieved from long-term memory (LTM) and combined, and the place where this combination occurs is called working memory. It consists of the phonological loop, which maintains limited information in an auditory code for a short time; the visuospatial sketchpad, which temporarily stores visual and spatial information; and the central executive, which integrates this information. Additionally, a revised working memory model has been proposed by adding the episodic buffer, which integrates episodic information stored in long-term memory into a single representation [10].

Working memory tests primarily utilize the reading span test and the visual recall test. The reading span test (RST), first devised by Daneman and Carpenter (1980), is also known as a working memory span task. It measures not just simple memory span but the overall capacity of working memory by requiring simultaneous storage and manipulation [11]. The visual recall test (VRT) involves presenting one or several shapes for a short duration (about 5–10 s) and then asking participants to recall and draw them immediately or after a certain period. This type of task is most commonly used, and paired-associate learning tests are also employed [12].

2.2. Cognitive Load Theory

Cognitive load generally refers to the amount of cognitive resources required during the process of learning or task resolution. For information to be learned, it must be processed in working memory. When the processing of information exceeds the capacity of working memory, leading to information being lost or overloaded, this indicates that the cognitive resources have been exceeded [13].

2.2.1. Dual Coding Theory

Paivio (1991) [14] explained that information entered through visual and auditory channels is processed in visual and auditory working memory, respectively, enabling effective cognitive activities. In the dual coding theory, when auditory and visual information are presented, they enter sensory memory through separate channels, with verbal information organized as auditory information and visual information as images. The organized information forms connections among them, which aid in learning; these are referred to as referential connections.

2.2.2. Cognitive Load in Image Processing

Understanding text and images involves integrating text and image information with prior knowledge stored in long-term memory [15]. Identifying whether text and images process and support two types of information resources is an important strategy for facilitating learning. When image information can only be understood in conjunction with text, comprehension of the text takes precedence [16]. In materials composed of images and text, when deriving meaning solely from images is challenging, the process of constructing meaning through text takes precedence [17].

2.2.3. Split-Attention Effect

Human cognitive capacity is limited, allowing only a very restricted amount of information to be processed in a short moment. When visual and auditory information are presented simultaneously, working memory processes them independently. However, when learning visual (images) and auditory (verbal) information simultaneously, if text, which is a type of visual information, is provided at the same time, it results in cognitive overload in the visual working memory. Providing both visual information and text at the same time causes an overload in the visual working memory capacity. This phenomenon of cognitive overload is known as the redundancy effect, which refers to the inability to effectively learn due to cognitive overload when too much information is presented. When visual information like text and images is provided at the same time, the learner’s visual attention is dispersed, leading to cognitive overload due to the limited capacity of human working memory [18].

2.3. Attention

In human information processing, key aspects include detection, filtering, and comprehension of stimuli. The attention process extracts, reduces, and selects important information, allocating cognitive resources to facilitate cognition and action execution. The capacity of working memory is a highly limited cognitive resource. Selective attention determines the priority of relatively important information, and without selectively placing important information in the working memory space, information overload can easily lead to task failure [19]. Measuring the attention allocated to stimuli, such as warning signs, can help identify the efficiency of the stimuli, with well-known attentional indicators like focused attention, selective attention, divided attention, and sustained attention being used for measurement [20]. Selective attention, divided attention, and sustained attention are closely related to hazardous stimuli. Selective attention is related to concentration and determining the priority of information, while divided attention relates to the distribution of cognitive resources among multiple cognitive tasks [21].

2.3.1. Selective Attention

Humans utilize the five senses to pay attention to different types of information, and among these, visual information plays the most significant role. The amount of information conveyed through visual attention is vast, but humans do not have the memory capacity to accommodate all cues received through visual attention. This selective process of cognitive processing is known as selective attention [22]. Selective attention and working memory have a special relationship, and working memory load decreases selective attention [23].

2.3.2. Divided Attention

Human information processing capacity is limited, and performance deteriorates when carrying out two or more tasks simultaneously as it is not possible to exceed the information processing capacity [21].

2.3.3. Eye Movement and Attention

Eye movement is not just the act of moving the eyeballs; it involves the optic nerve, which is connected to the brain. This optic nerve controls attention to store information as needed through eye movements, enhances recognition and recall of word lists, and influences the activation of cognitive abilities [24]. Attention to visual information naturally accompanies eye movement. This involves a process where, following visual stimuli, selective attention filters the information, allowing only the filtered information to be inputted into the brain’s storage. The cause of this selective attention can be attributed to the objectives pursued by the individual when viewing the target [25].

2.3.4. Attention Characteristics of Safety Signs

Warning signs play a crucial role in communicating hazardous information in the workplace. Their design characteristics, such as size, location, color, contrast, signal words, and pictograms, can significantly influence their attractiveness and encoding capabilities [26]. In this present study, four design factors were considered, location, color/contrast, signal words, and pictograms, to assess the visual attentiveness of construction workers. The study did not consider design size, as its focus was on the impact of multiple safety signs. To achieve this, multiple pictograms of a fixed size were posted in the same location. Research suggests that pictograms are highly effective in enhancing attention, improving encoding, and increasing comprehension [27,28]. They are especially useful in overcoming barriers such as language or illiteracy and can convey essential hazard factors, potential outcomes, and directive information at a glance [26]. Colors or other forms of contrast can also increase the visibility of warnings and the likelihood of information being encoded. They can even suggest various levels of risk, similar to signal words [27]. While warnings should generally be placed physically and temporally close to the hazard, their placement can be complex. Front-facing warnings, which allow for product identification, tend to attract more attention and are more likely to be noticed than warnings on the back or sides of the hazard. Additionally, warnings placed before instructions for performing tasks are more noticeable and likely to be encoded than those placed after the instructions [29].

The scope of the current research covers only the impact of multiple safety signs and design factors on the recall rate and visual attention, to examine the cognitive load limit of construction workers.

3. Methodology

Pictograms in safety signs are thought to attract attention and enhance encoding, generally being more effective in conveying information than text warnings. However, pictograms can lead to various interpretations among individuals and may cause significant cognitive load due to the large amount of information they carry. Construction sites contain various hazards, so safety signs are often posted simultaneously in multiple locations rather than individually. This study aims to analyze the effects of the number, arrangement, color, and design characteristics of multiple safety signs (pictogram only, text only, and pictogram with text), as well as the age, sex, and experience of workers, on recall rate and visual attention. The goal is to consider design characteristics for effective information transmission.

3.1. Experimental Subjects

The study was conducted with 513 Korean construction workers, aged between 20 and 50, with no perceptual problems, working at a construction site in Pyeongtaek (

Table 1. Test subjects.

Variable	Group	No. of People	Proportion (%)
Age group	20s	157	30.6
	30s	120	23.4
	40s	123	24.0
	50s	113	22.0
Career (year)	1 year less than	173	33.7
	1~2 year	50	9.7
	2~3 year	64	12.5
	3~4 year	34	6.6
	4~5 year	30	5.8
	5 year over	162	31.6
Sex	Male	478	93.2
	Female	35	6.8

). As of 2022, with a population of 1,740,000 workers in the construction industry [30], a minimum sample size of 385 is required for a confidence level of 95% and a margin of error of 5%.

3.2. Experimental Method

To examine the limits of working memory capacity for different design factors of multiple safety signs, three types of designs were created: type A (safety signs consisting only of warning text), type B (safety signs with both pictograms and warning text), and type C (safety signs consisting only of pictograms). Each design type was composed of 6, 9, and 12 multiple safety signs, as shown in Figure 2. The pictograms selected were those with high comprehensibility [31], including prohibition, warning, and directive signs, and used pictograms defined in ISO 7010 [32].

Prior the experiment, the participants were educated on the meanings and understanding the methods of safety sign pictograms. They were shown examples of multiple safety signs used on site, not for the purpose of testing their understanding of the training but to explain that their participation in the experiment was for the effective posting of safety signs.

To ensure understanding and willingness to participate in the study, a pilot experiment involving nine alphabets (Figure 3) was conducted prior to the commencement of the main experiment. In the analysis of the cognitive experiment results, responses from participants who did not participate in both the pilot and main experiment were excluded from the result compilation and analysis.

Commonly used safety signs on site are composed of pictograms and text, typically in a static form where the entire sign is exposed at once, rather than being dynamic or changing shape. Reflecting this common feature, participants were presented with multiple safety signs under a condition of full exposure.

To measure working memory capacity, the experiment was structured based on established methodologies such as the reading span test, which involves immediate recall of a list of words, and the visual recall test, which shows a few shapes for a short time (5–10 s) followed by an immediate recall task. For the visual recall test’s stimulus exposure time, the time it took for participants to reach the entrance from the point of first noticing a safety sign posted at the entrance was determined to be 7 s based on pre-interviews. Considering that workers on site do not make intentional efforts to memorize safety signs (like rehearsing) and that the retention time in short-term memory is within 20 s, participants were asked to recall and respond within 20 s immediately after a 7 s exposure.

Participants were shown the multiple safety signs on a TV monitor (1247.8 mm wide, 729.4 mm high) for 7 s. They were then instructed to freely draw or write on a pre-prepared grid-form response sheet within 20 s, recalling as much as they could in any form they remembered, whether in pictures or words.

4. Result

4.1. Cognitive Load of Multiple Safety Signs

4.1.1. Recall Number by Design Characteristics

In the study, regardless of the design characteristics of three types of signs—text only, pictogram with text, and pictogram only—over 95% of the participants recalled six or fewer items, and none recalled more than eight items. Specifically, for type B and type C signs, the number of items recalled decreased when more than nine items were shown. The average number of items recalled by participants was 3.32 (

Table 2. Number of recalls by design characteristic.

No. of Memory Types of Signs		1	2	3	4	5	6	7	8	9	10	11	12	Sum
Type A (text only)	12	1	9	32	31	25	13	1	1	0	0	0	0	113
		0.9%	8.0%	28.3%	27.4%	22.1%	11.5%	0.9%	0.9%
	9	3	19	36	29	20	5	5	1	0	0	0	0	118
		2.5%	16.1%	30.5%	24.6%	16.9%	4.2%	4.2%	0.8%
	6	6	28	53	18	7	1	0	0	0	0	0	0	113
		5.3%	24.8%	46.9%	15.9%	6.2%	0.9%
Type B (pictogram with text)	12	5	32	33	23	16	3	1	0	0	0	0	0	113
		4.4%	28.3%	29.2%	20.4%	14.2%	2.7%	0.9%
	9	6	14	48	29	18	3	0	0	0	0	0	0	118
		5.1%	11.9%	40.7%	24.6%	15.3%	2.5%
	6	5	24	43	25	8	8	0	0	0	0	0	0	113
		4.4%	21.2%	38.1%	22.1%	7.1%	7.1%
Type B (pictogram only)	12	7	37	36	29	5	3	0	0	0	0	0	0	117
		6.0%	31.6%	30.8%	24.8%	4.3%	2.6%
	9	7	37	52	45	16	6	2	0	0	0	0	0	165
		4.2%	22.4%	31.5%	27.3%	9.7%	3.6%	1.2%
	6	9	44	58	42	10	2	0	0	0	0	0	0	165
		5.5%	26.7%	35.2%	25.5%	6.1%	1.2%

, Figure 4).

4.1.2. Recall Number by Age and Experience

The average number of items recalled varied by age group, 3.68 items for those in their 20s, 3.36 items for those in their 30s, 3.15 items for those in their 40s, and 2.86 items for those in their 50s, showing a decrease in the average number of items recalled with increasing age (

Table 3. Average number of recalls by age.

Type of Signs Age		Text Only			Pictogram with Text			Pictogram Only			Grand Mean
		6	9	12	6	9	12	6	9	12
	20 s	3.33	4.03	4.39	3.67	3.89	3.72	3.18	3.75	3.16	3.68
	30 s	2.85	3.81	4.21	3.35	3.38	3.32	3.03	3.12	3.19	3.36
	40 s	2.71	3.69	3.67	2.96	3.31	2.58	3.07	3.40	3.00	3.15
	50 s	2.74	3.16	3.37	2.79	2.88	2.68	2.77	2.71	2.65	2.86
Grand mean		2.91	3.67	3.91	3.19	3.37	3.08	3.01	3.25	3.00

).

As for recall numbers by experience (

Table 4. Average number of recalls by work experience.

Type of Signs Age		Text Only			Pictogram with Text			Pictogram Only			Grand Mean
		6	9	12	6	9	12	6	9	12
	Less than 1	2.97	3.38	4.34	3.37	3.32	3.50	3.19	3.63	3.11	3.42
	1~2 year	2.93	3.93	3.79	3.21	3.43	3.07	2.92	3.46	3.40	3.35
	2~3 year	3.06	4.47	3.61	3.28	3.71	3.44	2.94	2.94	3.08	3.39
	3~4 year	3.11	3.88	4.78	3.44	3.88	3.67	3.22	3.67	2.75	3.60
	4~5 year	2.67	2.86	4.00	3.17	3.00	1.83	2.90	2.70	2.50	2.85
More than 5 year		2.89	3.74	3.68	3.14	3.31	2.79	2.93	3.05	2.86	3.15
Grand mean (year)		2.94	3.71	4.03	3.27	3.44	3.05	3.02	3.24	2.95

), there was no significant change in the number of items recalled as experience increased. Although higher educational levels in a group can lead to increased working memory capacity through learning and training, the increase in familiarity with safety signs due to increased experience of the workers did not lead to improved cognitive abilities.

4.2. Divided Attention in Multiple Safety Signs

The results of the experiment on the divided attention effect of multiple safety signs showed that for type B signs, the average number of items recalled was 3.27 from 6 signs, 3.41 from 9 signs, and 3.19 from 12 signs. For type C signs, the average number of items recalled was 3.04 from 6 signs, 3.32 from 9 signs, and 2.97 from 12 signs. No significant differences were found between the two design types, indicating that there was no increase in cognitive load due to the divided attention effect (Figure 5).

A normality test was conducted to determine whether the variables were normal (

Table 5. Normality test of recall count of pictogram and text, and pictogram.

	Kolmogorov–Smirnov		Shapiro–Wilk
	Stat. Value	Significance Probability	Stat. Value	Significance Probability
6	0.206	0.000	0.921	0.000
9	0.203	0.000	0.931	0.000
12	0.188	0.000	0.915	0.000

). The Shapiro–Wilk test, most commonly used for sample sizes between 3 and 50, was employed, and the Kolmogorov–Smirnov test was also conducted as a reference.

The Shapiro–Wilk test results indicated that values with a p-value less than 0.05 did not follow normal distribution. To investigate differences in variables across groups, the Mann–Whitney U test was used (

Table 6. Average and standard deviation of the number of recalls.

Variable	Items	Average	Standard Deviation	Z	p
6	Pictogram and text	3.27	1.212	−1.337	0.181
	Pictogram only	3.04	1.047
9	Pictogram and text	3.41	1.119	−0.919	0.358
	Pictogram only	3.32	1.219
12	Pictogram and text	3.19	1.257	−1.108	0.268
	Pictogram only	2.97	1.102

).

When there were six multiple safety signs, the average for type B was 3.27, and for type C, it was 3.04. With nine signs, the average for type B was 3.41, and for type C, it was 3.32. For 12 signs, the average for type B was 3.19, and for type C, it was 2.97. However, the Z-values and probability levels did not meet the criteria, thus failing to show significant results.

4.3. Cognitive Load of Pictograms and Text in Multiple Safety Signs

The results of the cognitive load experiment with pictograms and text in multiple safety signs are presented in Figure 6 and

Table 7. Normality test of recall count of pictogram and text.

	Kolmogorov–Smirnov		Shapiro–Wilk
	Stat. Value	Significance Probability	Stat. Value	Significance Probability
6	0.209	0.000	0.912	0.000
9	0.184	0.000	0.930	0.000
12	0.184	0.000	0.934	0.000

and

Table 8. Average and standard deviation of the number of recalls of pictogram and text.

Variable	Items	Average	Standard Deviation	Z	p
6	Pictogram only	3.04	1.047	−0.694	0.488
	Text only	2.96	0.976
9	Pictogram only	3.32	1.219	−2.248	0.025 *
	Text only	3.71	1.403
12	Pictogram only	2.97	1.102	−5.995	0.000 **
	Text only	4.01	1.278

* p < 0.05, ** p < 0.01.

. In type A multiple safety signs, the average number of items recalled was 2.96 for 6 signs, 3.71 for 9 signs, and 4.01 for 12 signs. For type C, the average numbers recalled were 2.96 for 6 signs, 3.32 for 9 signs, and 2.97 for 12 signs.

In multiple safety signs with fewer than six items, no significant difference was observed based on design characteristics. However, as the number of signs increased, the average number of items recalled was higher for text than for pictograms. Values with a p-value less than 0.05 in the Shapiro–Wilk test indicated non-normality. The Mann–Whitney U test was used to examine differences in variables by group.

Examining the cases with six items, the average for type C was 3.04, and for type A, it was 2.96. Looking at the test statistics, the Z-value was −0.694, and the probability level was 0.488, indicating that there was no statistically significant difference for six items. For nine items, the average for type C was 3.32, and for type A, it was 3.71. Examining the test statistics, the Z-value was −2.248, and the probability level was 0.025, indicating a statistically significant difference for nine items. Examining the cases with 12 items, the average for type C was 2.97, and for type A, it was 4.01. The test statistics showed a Z-value of −5.995 and a probability level of 0.000, indicating a statistically significant difference for 12 items. In the cases of 9 and 12 items, type A showed a statistically significantly higher average than type C.

4.4. Attention Characteristics of Multiple Safety Signs

4.4.1. Visual Attention Characteristics According to the Design Features of Multiple Safety Signs

The experimental results for the visual attention characteristics of multiple safety signs showed a sequential decrease in recall rate from left to right in the first row (

Table 9. Recall rate of multiple safety signs (text only, 12).

Column Row	First	Second	Third	Fourth
First	22.8%	19.9%	13.8%	8.3%
Second	13.8%	3.1%	2.8%	0.4%
Third	10.7%	2.6%	0.9%	0.9%

,

Table 10. Recall rate of multiple safety signs (text only, nine).

Column Row	First	Second	Third
First	26.0%	24.7%	18.9%
Second	13.7%	5.9%	3.7%
Third	2.1%	3.4%	1.6%

,

Table 11. Recall rate of multiple safety signs (pictogram and text, 12).

Column Row	First	Second	Third	Fourth
First	27.4%	17.3%	21.6%	15.1%
Second	4.9%	3.8%	4.4%	0.5%
Third	1.6%	1.6%	0.3%	1.4%

and

Table 12. Recall rate of multiple safety signs (pictogram and text, nine).

Column Row	First	Second	Third
First	27.1%	22.6%	20.4%
Second	9.2%	8.0%	4.7%
Third	4.0%	1.7%	2.2%

). Regardless of design characteristics, it was confirmed that selective attention was paid from the first presented cover.

4.4.2. Attention Characteristics Based on the Position and Color of Multiple Safety Signs

The selective attention characteristics based on the arrangement position and color of multiple safety signs were found to be more influenced by the position of placement than by the color of the pictograms (

Table 13. Recall rate of multiple safety signs (pictogram only, place by color).

Column Row	First	Second	Third
First	23.6%	18.5%	11.5%
Second	12.8%	4.4%	5.9%
Third	7.5%	9.1%	6.8%

and

Table 14. Recall rate of multiple safety signs (pictogram only, randomly place by color).

Column Row	First	Second	Third
First	22.8%	15.9%	17.1%
Second	13.6%	0.4%	7.7%
Third	3.9%	9.6%	10.0%

, Figure 7).

5. Discussion

The results of the study indicated that despite an increase in the amount of displayed information, participants were only able to recall an average of 3.32 items from multiple safety signs. Most participants, over 90%, were only able to recall between two and six items, which is consistent with Miller’s magic number 7 ± 2 and Luck and Vogel’s visual working memory capacity of three to four items. The familiarity workers have with their work environment and tasks can cause them to overlook safety signs, even when they are intentionally directed to pay attention to them. The results also showed that there was no significant difference in the number of items recalled based on age or experience, but there was a difference in the effectiveness of pictograms and warning texts in conveying information. Pictograms were found to have a higher cognitive load and were less effective in conveying information compared with warning texts, which was similar to the findings of Alphorns et al. (2003) [33], who recorded that the mean comprehension score of existing warning signs improved with the addition of text.

Furthermore, the result showed that presenting both pictograms and warning texts in multiple safety signs did not result in a divided attention effect. Although pictograms were highly noticeable, the warning texts were composed of short, easy-to-understand words and were spatially close to the pictograms, leading to referential connections and preventing the divided attention effect.

Additionally, the visual attention characteristics in multiple safety signs showed that to minimize cognitive load, the recall rate decreased sequentially from left to right, starting with the first stimulus presented. This indicated a stronger primacy effect than a regency effect.

Finally, selective attention in multiple safety signs, when size was consistent, was observed to be more influenced by the position of placement than by the color.

6. Conclusions

This study was conducted among 513 Korean construction workers aged between 20 and 50, who did not have any perceptual impairments, to explore the cognitive load limits of multiple safety signs present at construction sites, which inherently have various hazards.

More than 95% of the participants recalled between two and six signs, regardless of the design type (text, pictogram, or pictogram with text), with an average recall of 3.32 signs and none recalling more than 8 signs. It was found that the success rate of conveying hazard information decreased with the number of safety signs, 51.5% for 6 signs, 38.6% for 9 signs, and 28.5% for 12 signs, indicating that increasing the number of signs did not effectively communicate more information.

Additionally, it was observed that the number of safety signs recalled decreased with age, regardless of the education and career levels of the workers. While safety signs typically combine pictograms and text, it was most effective to use both for up to six signs. For communicating more than nine hazards, using only text became more effective as pictograms could be too complex and overloaded with information, and the effectiveness increased as the number of items increased.

The recall rate of multiple safety signs was more influenced by the arrangement order than by the color of the pictograms, decreasing from left to right starting from the first sign on the top row.

Based on this study, effective strategies for conveying information through multiple safety signs include the following:

• Compose the display with no more than three signs to effectively convey hazard information, considering the severity.
• Using both pictograms and text is most effective for up to six signs.
• For more than nine signs, using only text is advantageous.
• The perception of hazard information is more affected by age than by education or career, necessitating special attention for older workers.
• When multiple pieces of information must be provided, placing important information at the front is effective.

The limitations of the present study provide opportunities for future research to improve upon them. For instance, the study’s controlled experimental conditions and prior training of subjects may not fully reflect the actual work environment. In real-world settings, the effectiveness of safety signs may vary, and there may be no intentional efforts to remember them, leading to a lower rate of communication than the experimental results suggest. Furthermore, the study analyzed recall rates based on the visual attention characteristics of pictograms of fixed size. To further explore the factors that may affect attention characteristics, additional research using equipment such as eye trackers is needed to examine variables based on the size and color of the pictograms.