Eye-Tracking Advancements in Architecture: A Review of Recent Studies

Abstract

This Scoping Review (ScR) synthesizes advances in architectural eye-tracking (ET) research published between 2010 and 2024. Drawing on 75 peer-reviewed studies that met clear inclusion criteria, it monitors the field’s rapid expansion, from only 20 experiments before 2018, to more than 45 new investigations in the three years thereafter, situating these developments within the longer historical evolution of ET hardware and analytical paradigms. The review maps 13 recurrent areas of application, focusing on design evaluation, wayfinding and spatial navigation, end-user experience, and architectural education. Across these domains, ET reliably reveals where occupants focus, for how long, and in what sequence, providing objective evidence that complements designer intuition and conventional post-occupancy surveys. Experts and novices might display distinct gaze signatures; for example, architects spend longer fixating on contextual and structural cues, whereas lay users dwell on decorative details, highlighting possible pedagogical opportunities. Despite these benefits, persistent challenges include data loss in dynamic or outdoor settings, calibration drift, single-user hardware constraints, and the need to triangulate gaze metrics with cognitive or affective measures. Future research directions emphasize integrating ET with virtual or augmented reality (VR) (AR) to validate design interactively, improving mobile tracking accuracy, and establishing shared datasets to enable replication and meta-analysis. Overall, the study demonstrates that ET is maturing into an indispensable, evidence-based lens for creating more intuitive, legible, and human-centered architecture.

1. Introduction

The use of eye-tracking (ET) systems is important for architects, architecture researchers, and architecture students, insofar as they provide an objective and detailed comprehension into how individuals perceive and visually interact with built contexts [1]. These systems contribute to understanding how people visually explore architectonic contexts, identifying which elements attract most attention or go unnoticed. Such recognition helps assess whether design features such as circulation, lighting, or proportions are being perceived as intended, contributing to the efficiency, functionality and valorization of architecture projects [2].

A better understanding of how to utilize ET effectively in architectonic contexts, and of how to integrate ET into design knowledge, is required. This is the central problem this article seeks to address, as well as how to understand current limitations of ET and to advance our understanding of visual perception in architectonic contexts.

This article represents the most up-to-date review of ET in architectural research today, incorporating the latest studies through 2024 and providing a comprehensive overview comprising technological, methodological, and pedagogical aspects. In contrast to previous reviews, we integrate findings across diverse subfields to present a unified picture of how eye-tracking is transforming architecture research and practice.

In this review, we systematically examined recent ET studies in architecture (2010–2024) to understand how this technology has advanced architectural research and practice. We focus on key application areas (design evaluation, wayfinding, and education), summarize common findings, and identify limitations and future research needs. The aim of this Scoping Review (ScR) is to bridge ET data with design insights.

Building on [3], our first draft of this article did not have “a highly focused research question” as might be the case in a Systematic Review (SR). As a Scoping Study (ScS), it aimed “to identify all relevant literature regardless of study design”. With greater familiarity of the literature, we could then redefine search terms and “undertake more sensitive searches of the literature” in a second phase. As referred to in [3], “the process is not linear but iterative, requiring researchers to engage with each stage in a reflexive way”.

In our case, we started not with a precise question but with a general search based on key words: “architecture” and “eye-tracking”. In a second phase, we refined the search using Boolean terms, but we rejected these findings for reasons of practicality, as this resulted in too many articles for a single researcher. It was only in the third and final phase that we defined three exploratory research questions that we finally decided not to use, as this article evolved in a more structure-based form based on final theme-oriented literature research, combined with the findings from the first phase. This final theme-oriented research was suggested by a GenAI prospection. These three exploratory research questions were as follows: How can eye-tracking help improve architectural design? What are the key challenges in interpreting eye-tracking data for architecture? How does eye-tracking complement existing architectural methodologies?

Finally, in the third phase, corresponding to the elaboration of this article, the approach was changed to that of a ScR as it integrated a critical appreciation of some of the articles (see Section Challenges and Gaps).

As we did not use a clearly defined research question in this ScR research, we will therefore introduce the following PICO research question as a suggestion for its review: Does the use of eye-tracking technology, compared to traditional qualitative observation methods, provide architecture students, researchers, and practitioners more accurate, objective, and actionable insights into pedagogic visual perception, design evaluation, and spatial navigation within built environments?

1.1. Literature Search Strategy

This ScR was conducted using knowledge from the authors of this study based on research in the literature for a previous version of this review, corresponding to its first phase. We used the research strategy defined in [3] and searched for evidence on 27 November 2024 in Harzing’s Publish or Perish (Windows GUI Edition), version 8.17.4863.9118.

Inclusion criteria were:

• Publication in English.
• Full articles.
• Published in relevant scientific journals or conference proceedings.
• Published after 2010 and before 2024.
• Mentioning both “architecture” and “eye tracking”.

Exclusion criteria were:

• Not being written in English.
• Being just abstracts.
• Not referring to architecture.
• Repetitions.

We did not exclude articles based on quality scoring as our aim was to map all relevant literature; however, we note that study methodologies vary, and some limitations are discussed in Section Challenges and Gaps.

Our key term was “architecture, eye-tracking” across all searches. The review was a single-author effort over approximately 3 months.

The temporal scope of the analysis was 14 years (2010–2024). A previous classification was performed based on publication date: from the 75 articles selected, we found 29 (38.66%) articles published before 2021, 14 (18.66%) from 2022 and 2023, and 32 (42.66%) from 2024. The three Literature Reviews (LRs) found were from 2022, 2023, and 2024. The oldest of the articles chosen dated from 2015 (1 article) and from 2016 (1 article).

One of the LRs, from Aalto and Steinert [4], found that while only 20 experiments with ET in architecture had been conducted from 1976 until the end of 2018, even with a large variety of equipment and methods available, from 2019 to 2021 the field suddenly leaped forward with 46 new experimental studies in 3 years.

We analyzed all abstracts of the 75 articles selected, searching for objectives, methods, and results. This article theme classification method was based on checking the constituents of the articles, step by step in the order outlined below, until there could be no doubt as to their theme classification.

• Title.
• Abstract, highlights, and article aim.
• Concept(s) highlighted in the article (when existent).
• Research question(s) (when existent).
• Conclusion.
• Main article.
• Images, graphic(s), and table(s).

The different themes for classification were suggested by the articles themselves, with no a priori definition in an interactive process as [3]. In the end, we organized the main theme and subthemes by alphabetic order. We identified 13 main thematic categories of applications of ET in architecture, which included the focus of this article: architectural education and end-user experience. We also decided to include Vartanian et al. [5], due to its relevance to this ScR.

These thematic classifications were derived from the “intervention types” [3] and were validated by the literature sources of each of these themes.

As a possible Boolean keyword string for this literature search, we suggest, a posteriori, the following:

(“architect*” OR “architecture student*” OR “design professional*” OR “built environment researcher*”)

AND

(“eye-tracking” OR “eye tracking” OR “eye movement” OR “visual attention” OR “gaze tracking”)

AND

(“visual perception” OR “spatial navigation” OR “wayfinding” OR “design evaluation” OR “user experience” OR “architectural education” OR “design analysis”).

This ScR was conducted with reference to the PRISMA-ScR guidelines. A PRISMA flow diagram of study selection is provided in Figure 1, and the PRISMA-ScR checklist is available as Supplementary Materials.

Generative artificial intelligence (GenAI) has been used in this paper to generate the abstract; the text of some chapters and sections; tables, to assist in structuring study design; in the analysis and interpretation of data; and to fill out the PRISMA-ScR Checklist. To avoid bias and ensure analytical rigor, we carefully and critically revised the AI text. Our methodology in using AI gave preference to ChatGPT o3 deep research mode, uploading an anonymous draft of this article and using the prompt: “Could you discuss and conclude this article in a continuous text in APA style?” We sought to relate the article to Jan Gehl using the prompt, “Could you write one or two paragraphs to add to the conclusion connecting this study to Jan Gehl’s investigation?”, allowing ChatGPT to add whatever important information it might find and then critically reviewing this. Finally, we asked for a Buildings-orientated critique of the article and used it to introduce improvements in the article, using the prompt, “Could you make a critique of this article, considering Buildings Journal (https://www.mdpi.com/journal/buildings) (accessed on 20 May 2025)?”

As shown in Figure 2, we start this article by presenting a short history of ET, which allows us to first have a general view of this technology and its functionality as a tool to perceive our exterior visual environment.

1.2. Evolution of Eye-Tracking

ET concerns the where, what, and how, when we are focused on a determined visual stimulus. The focal points can be a painting, a photograph, a sculpture, or, as already mentioned, an architectonic context. Thus, ET can allow us to glimpse how individuals visually engage with art. Moreover, since the early 1990s, with ET becoming more widely accessible, this tool has generated great interest in the user experience field (UX) [6].

Around 2002, according to Duchowski [7], ET was arriving to its fourth era with the emergence of interactive applications (the first three eras are clearly summarized by Rayner [8] and cited by Duchowski [7]). In the first era, starting in 1879 and lasting until 1920, French ophthalmologist Louis Émile Javal observed eye movement during silent reading, coining the term saccade. At this time, many basic facts about eye movement were discovered, including saccade suppression (blindness during a saccade), saccade latency (time spent to plan and execute a saccade), and the measure of our amplitude of vision. The second era coincided with the behaviorist movement in experimental psychology and had a more applied focus, with less research emphasis on eye movement theory. Finally, the third era, which began in the mid-1970s, was marked by progress in eye movement recording systems.

On the other hand, more concerned with ET hardware, according to Holmqvist and Andersson [2], during the first era, Javal, Delabarre, Dodge, Buswell, Yarbus, and others built mechanical and optical hardware, recording analog data which was analyzed by hand. In the second era, beginning in the 1970s, researchers built electrical hardware and computer software for recording and analyzing analog and digital data. Finally, in 2017 we entered a third era. By then, eye trackers had become a commercial product sold or lent to researchers, who could use prepackaged software from manufacturers to easily record and process data.

Notably, Duchowski [8] and Holmqvist and Andersson [2] had different understandings of periods of ET history. Duchowski [8] adopts a methodological approach that classifies historical phases predominantly by the analytical and research paradigms that emerged in the different phases. Rather than focusing strictly on hardware, this author highlights shifting theoretical perspectives and experimental designs. In contrast, Holmqvist and Andersson’s [2] overview of ET history emphasizes the technological progression of recording and measuring ocular movement, segmenting developments based on significant developments in ET equipment. However, these two authors place particular focus on how hardware progresses, from rudimentary mechanical devices to modern video-based systems, shaped the evolution of this field. Holmqvist and Andersson [2] view the different periods through the ability of the technology to capture increasingly precise eye movement data, while Duchowski [8] defines similar historical periods through a prism of the methodologies and theoretical frameworks that arose, sometimes in response to technological advances.

Various authors [1,2,9] point out that the main method used today for the capture of gaze movements, possibly the most widely applied ET technique, is video-based corneal reflection. The first use of this method dates to 1901 [10]. Techniques were developed in the 1950s using contact lenses attached to devices such as small mirrors and coils of wire. Subsequently, comprehensive data was collected through devices measuring in physical contact with the eyeball. However, these late systems were very intrusive. The video-based corneal reflection systems are accurate and reliable, non-intrusive, and are now commercially available.

After this quick summary of ET history, it is important to focus on what ET is capable of studying in terms of the Human Visual System (HVS), specifically, the neurology which underpins our gaze and gaze behavior.

1.3. The Human Visual System

According to Holmqvist and Andersson [2], the human eye lets light enter through the pupil. This eye turns an exterior image upside down and projects it onto the anterior face of the eyeball, the retina. The retina is covered with light sensitive cells, named cones and rods. These cells transform light into electrical signals, which are sent through the optic nerve to the visual cortex in the brain. While cones are sensitive to visual detail and provide color vision, rods support vision under dim lighting conditions. In the retina, we find a small area called fovea. The fovea encompasses less than 2° of the visual field, and we only have high-acuity vision in this short angle. To see a selected object sharply, we must move our eyes. This foveal information is prioritized when processed to the brain.

Still, according to Holmqvist and Andersson [2], the most common event reported through ET does not relate to a movement, but a state when the eye remains immobile over a period. It is called fixation and lasts from tens of milliseconds (ms) up to several seconds. However, in this position, the eye is not completely stopped but moves. These movements are called tremor, microsaccades, and drifts [2,11].

These same authors, Holmqvist and Andersson [2], explain that a tremor is a small movement, of which the exact role is unclear, and which is often due to imprecise muscle control, while drifts are slow movements which take the eye from its focus, and microsaccades compensate these drifts by quickly bringing the eye back to its original position.

Finally, Holmqvist and Andersson [2] mention that the quick eye motion from one fixation to another is called saccade. Saccades are very fast and take typically 30–80 ms to complete. Moreover, many eye movement paradigms assume that we are blind during most of the saccade. Saccades rarely take the shortest path between two points but can undergo one or several shapes and curvatures. Mostly, saccades do not stop directly at the intended target but continue to wobble a little before coming to stop.

1.4. Types of Eye-Trackers

According to Duchowski, the ET systems available in 2017 were primarily head-mounted or table-mounted. Head-mounted devices were often integrated into portable glasses, allowing greater mobility for conducting experiments outside in controlled contexts. Table-mounted devices were usually affixed to a computer screen, and are better suited for static, laboratory-based studies. Holmqvist and Andersson [2] also categorize these systems by identifying eye-mounted eye-trackers, head-mounted eye-trackers, tower-mounted eye-trackers, and remote eye-trackers, which collectively illustrate the diverse technological approaches researchers employ to capture visual attention data.

Although, according to Holmqvist and Andersson [2], mounting an eye-tracker on to participant eyes can present several problems, it still offers the possibility of high-quality data. Hence, head-mounted eye-trackers place active parts on the head of the participant, on a helmet, cap, or a pair of glasses. Simultaneously, a scene camera records the stimulus. These head-mounted eye-trackers allow the participant maximum mobility, especially if the recording computer is small and lightweight. However, it is important to remark that the camera angle towards the eye can, in principle, be shifted and adapted to the individual participant and specific task. In addition, we can find other human interfaces such as fMRI eye-trackers and even primate eye-trackers. These last interfaces, and virtual reality (VR) eye-trackers, are all versions of mobile eye-trackers, and the stimulus display is fixed in relation to the camera and the head of the participant. However, while 3D objects are not fixed and the data will show a variable level of convergence due to depth, a constant level of accommodation will simultaneously take place.

After this presentation of ET history, HVS, and types of eye-trackers, we will examine the importance of human visual perception in architecture and how this understanding can help architects, researchers, and students of architecture in an innovative way, bringing new contributions to projects, research, and learning.

1.5. The Importance of Studying Visual Perception in Architecture

Several architectural features influence the way we experience architectonic contexts. Therefore, how we visually perceive architecture is imperative either to find our way throughout these architectonic contexts or to find its beauty.

Among several authors, namely the founding study by Ulrich [12], cited by Wiener and Franz [13], contextual characteristics of architecture influence subjective experience. Several theories thus explain human behavior and experience through the interdependency of individuals with their environment.

One author, O’Neill [14], demonstrated that wayfinding performance decreased with increasing plan complexity. In 2015, a few theories and empirical studies aimed at analyzing this interdependency objectively using ET (notwithstanding the researchers making use of qualitative descriptions for a few selected spatial situations).

Other authors argue that certain architectural elements like ceiling height have a strong effect on individuals [5], or that the perceived degree of movement freedom through an architectonic context, i.e., perceived enclosure, can have an impact on beauty judgments concerning architecture, together with decisions to enter or exit those contexts. Vartanian cites Appleton [15] to make the point that judgment of an architectonic context as esthetically pleasing results from its inclusion of certain characteristics (e.g., shapes, colors, spatial arrangements), suggesting favorability for survival (namely, “to see without being seen” and “to be not seen”).

Now that we are aware of the importance of studying visual perception in architecture, we will check the history and future perspectives of ET use in architecture. These usages of ET in architecture are still developing, as we will see, and ET is being increasingly applied in architecture, mainly by academic researchers. However, these usages should be extended to architects and students, as proposed by several scholars.

1.6. Eye-Tracking Architectural Applications, History, and Future Perspectives

The pioneers of ET applications to architecture are both [16,17]. According to Aalto and Steinert [4], Buswell [16] included images of architecture in his pioneering study, and Yarbus [16] published the earliest guide to ET research. Both these studies inspired Janssens [18] to publish “an exceptionally thorough study that combines ET with verbal descriptions and semantic ratings to examine the pleasantness and ease of identification of 10 types of Swedish building exteriors” [4]. According to these last authors, this work [18] might be the first use of ET in architecture. This same author, Janssens [19], has noted that architects and non-architects might see the world differently, but, as Aalto and Steinert [4] suggest, this author’s most important contribution might be highlighting the relevance of ET for architectural research.

Some researchers such as Weber et al. [20] have suggested that objects are seen differently than mere contours. Later, other authors, such as Foulsham et al. [21], indicate that a controlled laboratorial experiment can provide more significant quantifiable results about architecture than interviews or surveys; important results for a better understanding of the individual factors affecting experimental findings.

Aalto and Steinert [4] claim that up to 2021 ET had been used in architecture in multiple ways, such as remembering visual experience [22], visual preference [23], viewing objects and elements [24], architectural experience [25], creative performance [26], validity in building heritage and degradation [27,28], design process [29], triangulation [30], and education [28].

Finally, Aalto and Steinert [4] recommend the following for future research on the application of ET for the analysis of architecture:

• Use of VR to systematically verify designs in development to ensure that the gaze patterns of the users match what we want to be read in an architectonic context.
• Develop ET methodology, methods, and triangulation to deal with the complexity or in situ experiments.
• Further develop best practices for laboratory-based experiments.
• Use of replication and verification studies on analysis of gaze differences between architects and non-architects, on the relationship between gaze patterns and preferences, and on the role of individual building elements in attracting the gaze.
• Developing further new and rapid approaches such as visual attention software (VAS).

Having shortly reviewed the history of ET architectural applications and mentioned future perspectives for this field, we conclude this introduction. In the next chapter, we will broaden our knowledge about the role of ET in understanding architecture.

2. Understanding Eye-Tracking in Architecture

2.1. Use and Technical Principles

In architectural research, different types of ET hardware enable architects, researchers, and students of architecture to study how individuals visually engage with physical or virtual designs, thus revealing how building users perceive space, respond to environmental cues, and navigate complex structures (see Figure 3) [31,32,33]. By selecting the ET system that fits the required mobility and experimental control, either a head-mounted system for in situ observations by pedestrians inside a building, or a table-mounted system for assessing the gaze patterns of users on digital architectural renderings, researchers can produce targeted insights to inform user-centered design strategies or improve spatial experiences.

After discussing how ET can help architectural research, outlining some ET technical principles in architecture, we will refer and define eye movement concerning architectural research. We will go through Henderson and Hollingworth’s [34] exploration of scenes, view, and perception, relating these ways of seeing scenes to the viewing of architectonic contexts.

2.2. Fixations, Saccades, and Scan Paths in Architecture

According to Henderson and Hollingworth [34], in the literature, scene is typically defined as a view of a real-world context comprising “background elements and multiple discrete objects arranged in a spatially licensed manner. Background elements are taken to be larger-scale, immovable surfaces and structures, such as ground, walls, floors, and mountains, whereas objects are smaller-scale discrete entities that are manipulable (e.g., can be moved) within a scene”. The distinction between a scene and an object is dependent on the spatial scale. However, as it might be difficult to determine the approximate scale of a scene, most researchers of scene perception use a human scale. Henderson and Hollingworth [34] have done the same.

According to these authors, there are three levels of seeing when we first perceive a scene: low-level or early vision, intermediate-level, and high-level. In low-level, we extract the physical properties, depth, color, and texture, and generate representations of surfaces and edges [35]. On intermediate level, we extract shape and determine spatial relations [36]. Finally, in high-level vision, we map from visual representations to meaning, which is followed by an active acquisition of information stored in short-term memory, and the identification of objects and scenes. We will analyze this last level of seeing based on Henderson and Hollingworth [34].

During eye movement in scene perception, these authors conclude that, during scene viewing, fixation positions are non-random, with visually and semantically informative regions clustering these gaze positions. This attraction based on meaning is not immediate, but eyes spotting these regions may dwell there. For architecture viewing, this conclusion might imply that our gaze lingers and returns to those architectural features with most meaning in an architectonic context.

Concerning scene representation retained across a saccade, these authors conclude that during complex, natural scene viewing, only a limited amount of information is carried across saccades. And this information is coded and stored in an abstract (nonperceptual) format. Moreover, our experience of a complete and integrated visual world results from an illusion or construction based on “an abstract conceptual representation coding general information about the scene (e.g., its category) combined with perceptual information derived from the current fixation” [34]. For architecture viewing, this conclusion might imply that when we move our gaze through an architectonic context, an abstract representation of this context is developed on the basis of our previous conceptual information about this scene, mingled with what we are fixating on in a given moment.

Finally, concerning object and scene identification, the authors conclude that a consistent scene facilitates the identification of objects, despite the existence of methodological problems found in the studies [37,38] analyzed by Henderson and Hollingworth [34]. The authors also find that there are advantages for the identification of consistent versus inconsistent objects, though more recent studies have not found this advantage [39]. The functional isolation model, which proposes that object identification is isolated from expectations derived from scene knowledge, provided the best explanation for this problem. For architecture viewing, this conclusion might imply that expectations imputed by viewing architectonic contexts do not influence the identification of architectural features and that consistent architectonic scenes are better identified.

In the following section, we will present advantages and limitations in the use of ET in architecture. Although the use of ET presents many advantages, there are still important challenges ahead.

2.3. Advantages and Limitations of the Use of Eye-Tracking in Architecture

Mahmoud et al. [40], in their LR, reference several advantages and disadvantages related to the use of ET in architectural research.

As advantages:

• The interest in the experimental side of architectural research is increased.
• ET provides a graphically and numerically accurate communication of viewer experience.
• The acceptance of design orders which require in-depth analysis of the visual needs of the users is further guaranteed.
• The architectural importance of visual perception and of proximity to users are highlighted.

Whereas, as disadvantages:

• Maintenance and conservation of ET hardware is complex.
• ET equipment is not available for public use.
• ET devices are for only one user at a time.
• ET tools often still lack latency and accuracy.

These authors, Mahmoud et al. [40], reviewed five ET studies: [24,41,42,43,44].

One year later, Zhipeng and Pesarakli [45] reviewed 50 articles on ET and environmental research, including, ones specifically related to architecture and interior spaces: 10 studies on “examining the effects of environmental features (e.g., lighting, color, spatial arrangement, material and outdoor view) on people’s behavioral, emotional, and physiological responses”; 3 studies on “evaluating the effects of different types of stimuli (e.g., sketch, image, 3D representation and VR) on eye behaviors”; and 7 studies on “assessing the influences of environmental features on people’s indoor wayfinding performance and behaviors”.

Concerning these 50 studies, Zhipeng and Pesarakli [45] found several articles discussing the limitations of ET:

• Interpretations of gaze behaviors, fixations, and scan paths might not provide direct information on the brain activities (e.g., emotion, cognition, and attention) of the subject; found in the article by [46].
• Some limitations concerning ET glasses: “Eye-tracking became challenging when subjects had free movement, which incurred an inaccurate estimate of head direction/position and gaze direction in 3D coordinates” [45]; found in the article by [47].
• “Data loss was a common problem for outdoor mobile eye-tracking, possibly due to bright lighting conditions and head/body movements” [45]; found in the articles by [48,49].
• “Inaccuracy between the captured and the actual gaze points” [45]; found in the article by [50].
• Some studies used pupil size as an indicator of emotional state. “However, the pupil size and emotional arousal relationship was complex, and the pupil size was also influenced by other factors such as cognitive processing load” [45]; found in the articles by [51,52,53,54]. Factors like “light quantity and contrast” also influence pupil size [45]; found in the articles by [55,56].
• Concerning data analysis, for studies using mobile ET devices, it took significant time and effort to code and analyze visual behaviors, as the views of the subject constantly changed during the experiment [45]; found in the articles by [57,58].

As we see, although there seem to be some important advantages, the use of ET in architectural studies is still a major challenge. Now we will set our attention to the role of ET in architectural research, where ET seems to perform multiple tasks: education, end-user satisfaction, wayfinding, and understanding architectonic contexts features, among others (see Figure 4).

3. Role of Eye-Tracking in Architectural Education and Research

3.1. Eye-Tracking Applications in Architecture Design Pedagogy

After four years of cooperation and research, Rusnak and Rabiega [28] published an article concluding that they are certain of the importance of ET for architects and the future education of urban planners, highlighting how ET makes it possible to record what attracts or distracts the attention of the users of specific spaces, as well as how elements defining squares, streets, and passages are perceived by people, improving orientation in the city and inside buildings. Moreover, ET makes it possible to verify expectations of architectural projects and promote the self-improvement and self-development of students and teachers, helping them gain knowledge in experimental research.

Rusnak and Rabiega [28] enumerated several strong arguments for introducing ET in the curricula of architecture learning:

• It is an inventive technique to guide the attention of future architects to the topic of order in architecture and urban planning, broadening their knowledge on the perception of architecture, i.e., how to attract the gaze of the users of architecture through design, while at the same time appropriately inscribing the project of someone in the natural or historical context.
• Increases the interest of students in the experimental side of research in architecture, which may lead to solving architectural projects with more creativity.
• Broadens the social and technological skills of students, which may facilitate their future acceptance of non-standard and complex architectural project orders that require in-depth analysis of the visual requirements of the users, as well as interdisciplinary cooperation.
• Self-monitoring of both teachers and students.
• Positively influences student–teacher working relationships, which may facilitate progress to advanced studies, e.g., master’s and doctorates.
• Promotes the academic institution, distinguishing it from other research centers, both due to these advanced technological solutions and by adjusting learning requirements to real needs of the users.
• Educates the architectural public by interesting them in the buildings they see day-to-day and promoting the profession of architects.

Nevertheless, Rusnak and Rabiega [28] also found a few problems:

• High cost of ET purchases, maintenance, conservation, and insurance.
• Necessity of a room for use for up to 12 persons and, depending on the experience desired, possibly a laboratory.
• Teachers may contest the legitimate use of ET for self-analysis, as it requires adding work hours and extra effort as well as an open-minded and self-critical approach.
• Classes need to be held in-person to manipulate ET, which would be impossible under exceptional conditions like COVID.

While Rusnak and Rabiega [28] emphasized that these advantages and disadvantages require verification, the literature on the use of ET in other fields seems to confirm the usefulness of ET in architecture education.

After seeing advantages and disadvantages of using ET in architecture education, we will see how these devices suggest a more precise idea of what is an expert and novice gaze in architecture by means of examining how ET behaves differently in these two cases.

3.2. Differentiating Expert and Novice Gaze Behavior in Architectural Education

While we might assume that teachers and students look at architecture in different ways, ET should allow us to identify where, what, and how they each appreciate an architectonic context. Whether this difference might be of pedagogic value is also of interest.

In their article about differences between the expert and non-expert gaze, Jam et al. [59] refer to numerous studies in the literature exploring these perceptual differences in architecture and urban planning [60,61,62,63].

Jam et al. [59] investigated “the impact of expertise on preference, visual exploration, and cognitive load experienced during the esthetic judgment of façades” in Theeran, using a psychophysical paradigm and ET to test four hypotheses. They found significant differences in how experts viewed façades versus non-experts: experts had longer fixations and saccades and paid more attention to context and structure, whereas non-experts had more fixations on decorative elements and shorter scan paths. They also noted that experts showed signs of higher cognitive load, suggesting deeper processing.

Their investigation concludes by stating that the analyzed visual esthetic evaluation of the façades could be influenced by their physical elements and attributes, and a personal cognitive variable such as expertise. They close by recommending further investigation using ET.

We believe that students of architecture are also non-experts in architecture, but with a higher level of expertise than the individuals participating in this study by Jam et al. [59], whose research suggests that we would also expect to find different ways of viewing architecture between teachers and students of architecture by using ET. Consequently, this tool might be valuable in the teaching of architecture, guiding students to expert ways of seeing.

In the next section, we will see how it is possible and advisable to use ET to improve wayfinding through architectonic contexts. Consequently, user-end experiences of architecture contexts may be a means of using ET for enhancing architectural design.

3.3. Evaluating Wayfinding and Spatial Navigation in Architectural Research

ET can operate as a robust tool for understanding how people navigate and orient themselves within architectonic contexts, offering valuable data about the gaze of users and attentional priorities during wayfinding tasks [13]. By correlating ET metrics such as fixation density and saccadic patterns with architectural features, researchers can reveal which environmental elements users rely on when creating mental representations of the layout of a building. For instance, individuals might repeatedly fixate on “anchor points” or salient features like doorways, windows, or signage, which facilitate orientation and route selection [13].

According to Sun, Li, Lin, and Hu [31], places where visitors have a clear wayfinding purpose such as transportation facilities are important in wayfinding research, focusing on the importance of efficient signage systems to improving wayfinding in airports. Wayfinding is also very important in railway stations, and Zeng, Zhang, and Zhang [32] find that “connectivity and visual field area of wayfinding nodes have strong positive correlation with passengers”.

Wayfinding is a process that moves from environmental perception to decision-making and, through ET, specific causes of a wrong wayfinding decision can be rigorously determined and analyzed [31]. Current research on architectural wayfinding uses a variety of representation methods, including field reality scenes, three-dimensional VR models, panoramic photographs, etc., and research is focused on two broad areas: guide signs and space patterns [32].

According to Wu, Chen, Zhao, and Xue [64], wayfinding is a continuous process involving perception, decision-making, and execution, and wayfinding studies normally choose to employ field reality scene or VR methods.

Wayfinding in architectonic and urban contexts is therefore an important field for ET use, with this technology being crucial for implementation of good practices. In the following section, we will analyze the attention of the viewer and perceptual hierarchy when exploring interior spaces, mainly using a study by Vartanian et al. [5].

3.4. Analyzing Visual Hierarchy and Attention Distribution in Interior Spaces in End-User Experience Research

An article by Vartanian et al. [5] about ET provides valuable insights concerning how specific architectural features, such as ceiling height or enclosure, can condition the attention of the viewer. This study is also concerned with perceptual hierarchy when exploring interior spaces. By recording fixations and saccadic movement, researchers can pinpoint which elements, like high or low ceilings, open corridors, or enclosed corners, attract immediate attention. Vartanian et al. [5] verified that the judgments of people on beauty and comfort are closely tied to the spatial arrangement they perceive, highlighting that ceiling height and enclosure levels serve as major cues during the scanning of the viewers of an architectonic context.

Moreover, visual hierarchy revealed through ET often shows that individuals assign more time to areas they find crucial for orientation, doorways, functional furnishings, or lighting sources, and to features that make a room feel expansive or constricted [5]. In open, airy rooms with higher ceilings, participants typically exhibit fewer re-checks or back-and-forth saccades to reassure themselves of the spatial dimensions. By contrast, more enclosed or lower-ceilinged designs provoke longer and more frequent fixations on potential congested areas, presumably due to higher perceived complexity [5].

Together, these patterns show how the interaction of enclosure and ceiling height influences attentional distribution and comfort levels of users. Finally, architects can use these ET insights to conceive interior architectonic contexts with effective direct visual interest, reinforce intended pathways, and promote a sense of esthetic harmony [5].

After analyzing the use of ET for interior features of architectonic contexts, in the next sections, we shall discuss and conclude the findings of this study.

4. Discussion

The findings of this study underscore the status of ET as a powerful lens for understanding visual perception in architecture, revealing how people explore and experience architectonic contexts. Over recent decades, advances in ET technology and methodology have made it possible to gather precise, real-time data on where viewers focus in architectonic contexts [1].

As a result, researchers can now objectively document attention patterns in spaces ranging from buildings and streets to interior rooms. Indeed, ET is increasingly applied in architectural research, with Aalto and Steinert [4] reporting that after only approximately 20 studies from 1976 to 2018, there were 46 new ET-based experiments in architecture between 2019 and 2021. This growing body of work also reveals the value of ET in connecting design intent and actual user experience.

By capturing how elements like lighting, layout, or forms attract (or fail to attract) attention, ET provides evidence-based insights that enrich our understanding of the HVS in context. It confirms, for example, that observers do not passively see a space all at once; rather, vision is an active process of selection in the environment, guided by both the physical features present and the goals and expectations of the viewers [34]. In architectonic scenes, this might mean that people tend to sequentially focus on spatial cues that help them make sense of the environment, a process that ET can elucidate in detail.

One prominent theme that emerged in this study is the difference in gaze behavior between experts (e.g., trained architects) and novices when observing architecture. This ScR and analysis build on previous research, finding that expertise significantly shapes visual scanning patterns. In a study with some limitations, Jam et al. [59] showed that architects tend to have larger and more structured scan paths when evaluating building façades, whereas non-architects tend to fixate more on immediate, decorative details. In the study of this authors, experts more frequently scrutinized contextual and structural elements of façades (overall form or integration with surroundings), and showed fewer fixations on ornamental features, compared to novices. This might suggest that with training, architects develop schemas or expectations that guide their eyes efficiently toward functionally or compositionally relevant aspects of a design, rather than getting distracted by surface ornamentation. Experts were also perceived to be “more active and attentive” in their viewing, indicating they cover a scene with purposeful eye movement to extract information. Such discrepancies in visual strategy are not merely academic but rather suggest important implications for how we teach and practice design. In any case, further research in the field is recommended.

In education, making students aware of these expert–novice differences can be very instructive. If novice students can learn where to look, for instance, to pay more attention to contextual cues or spatial organization, as experts do, they may improve their ability to evaluate and create designs. ET can facilitate this pedagogical goal by providing immediate, objective feedback. For example, during a critique, a teacher might use ET to show a student which key features of a building they overlooked, or conversely how a gaze of an expert lingered on areas the student underappreciated. Indeed, scholars have begun to argue that ET is a valuable tool for design studios, allowing them to “guide the attention of future architects” towards more expert-like observation patterns. Rusnak and Rabiega [28] stress that incorporating ET into architecture curricula can broaden the understanding of students on how designs are perceived, enhance their appreciation of user experience, and even promote greater self-reflection in both students and teachers.

However, benefits for education come with practical limitations. As Rusnak and Rabiega [28] remark, high equipment costs and the requirement for dedicated laboratory space or hardware can pose challenges to schools. Additionally, teachers must be trained to interpret ET data and integrate it meaningfully into the design critique process. Despite these difficulties, the consensus emerging from the literature is that the pedagogical return of ET, in promoting evidence-based design learning and cherishing more attentive, user-conscious architects, is significant.

Another central contribution of ET research to architecture is user-centered design practice, particularly in areas like spatial navigation, wayfinding, and interior layout optimization. ET offers extraordinary insights into how people navigate through spaces and which features they rely on for orientation. By analyzing the gaze paths of the users and fixation clusters, architects can identify environmental elements that attract attention and operate as cognitive “anchor points”. Studies in complex public buildings have found that users naturally fixate on doors, signage, and distinctive markers to orientate themselves. If an expected cue (a directional sign) is constantly missed in gaze data, it might be a clear suggestion that the design or placement of this element is inadequate.

Empirical research further emphasizes this utility. In an airport wayfinding study, Sun et al. [31] used ET to show that efficient navigation is sustained in well-designed signage systems, as the eyes of the participants are attracted to signs at decision points. Similarly, Zeng et al. [32] found in a railway station context that the “connectivity and visual field area” of key wayfinding nodes (e.g., junctions, exits) had a strong positive correlation with how easily passengers found their way.

Besides wayfinding, ET research into interior space perception shows how design influences both attention and emotional responses. Notably, Vartanian et al. [5], on the basis of gaze behavior, demonstrated that ceiling height and enclosure affect not just esthetic judgments but decisions about approach and avoidance. Participants in more open, high-ceiling rooms exhibited more relaxed viewing patterns, fewer back-and-forth fixations, less visual checking of the environment, and were more inclined to classify such spaces as beautiful and inviting. In contrast, low-ceiling or more enclosed rooms elicited denser fixation patterns (people visually “searched” the space more thoroughly) and were more frequently related with the desire to exit the space. For architects, this implies that subtle design choices, which change how space is visually perceived (spacious vs. confined, open views vs. obstructed views), can be objectively correlated with user comfort and behavior.

Finally, the advantages of ET for insights being clear, our discussion also brings to light certain methodological and technological limitations. Firstly, the probability of bias in interpretation. ET records where someone looks, but not why. The same gaze pattern can have multiple explanations, and ET alone cannot distinguish between a prolonged fascinated gaze and another driven by confusion or discomfort. As Zhipeng and Pesarakli [45] emphasize in their review, metrics like fixations and scan paths “might not provide direct information” about the underlying cognitive or emotional state of a viewer. In research, this might mean that while we can detect that a participant stared for 5 s at a particular painting or façade detail, complementary methods (such as interviews, think-aloud protocols, questionnaires like the kansei engineering [65] method, or physiological measures from biosensors) are needed to ascertain whether that gaze indicates esthetic appreciation, puzzlement, or something else. Without such context, there is a risk of misinterpreting the data.

Other limitations concern hardware and data collection constraints. Modern eye trackers have evolved substantially from the difficult setups of the past, becoming relatively non-intrusive and user-friendly, yet they still have limitations. Mobile head-mounted eye trackers (ET glasses) introduce complexity when users are free to move in a real environment. Even a small head movement can introduce a mistake in mapping the gaze of a 3D space, leading to reduced accuracy in determining exactly what object was viewed.

As multiple studies have reported, outdoor usage adds further difficulty, as bright lighting and wide head motions often cause data loss or tracking dropouts. In our study context, if an architect were to use ET glasses to study how people explore an urban plaza, he might encounter segments of missing data whenever a user quickly turns their head or walks under direct sunlight, potentially leaving gaps in the visual record. Additionally, calibration drift over time can mean that the gaze point recorded might be slightly offset from the true gaze, requiring careful post-validation. This calibration drift might be tackled through the technological improvement of software and hardware.

Even in controlled settings, current commercial eye trackers still have finite selection rates and precision. Mahmoud et al. [40] note that ET tools today “still [lack] latency and accuracy” to some degree. This can be critical when analyzing fast eye movement or very small details. A slight timing lag or positional error might cloud the analysis of whether a brief glance fell on a particular sign or just next to it.

Another practical limitation is that ET systems just record one person at a time, which makes large-group studies or collaborative scenario tracking impossible. In research, we might be interested in social dynamics (e.g., how people in a crowd collectively attend to a public art piece or how two people talking in a space look around). High costs and setup requirements further constrain widespread usage; setting up an ET laboratory for architecture may require significant investment in equipment and space, which not all firms or schools can afford.

Despite these challenges, it is important to recognize that different types of ET setups offer a spectrum of options, and selecting the right setup can mitigate some limitations depending on the research or design question. A steady, screen-based eye tracker (often mounted below a computer monitor) is particularly useful in studios where we utilize images, plans, or VR scenes on a screen under controlled lighting. Such systems usually possess high precision and stability, making them ideal for pinpointing minute gaze differences when comparing design alternatives or evaluating visual attention on drawings. However, they inherently restrict the movement of the participants and field of view, thus sacrificing realism, as a person viewing a building on a monitor may not behave exactly as they would in situ. On the other hand, head-mounted mobile eye trackers allow users to walk through real or mock-up architectonic contexts, providing rich data on how attention evolves naturally in space, but at the cost of lower spatial accuracy and more complex data processing. Our review noticed that analyzing data from mobile ET can be labor-intensive, as a scene is continuously changing, and each fixation must be mapped to a dynamic reference frame. Still, for studies of wayfinding or immersive experience, the mobile approach is the best of options.

An emerging middle ground is the use of VR headsets with incorporated ET. This system can immerse participants in a full-scale 3D simulation while recording their gaze, combining some benefits of real-world immersion with the experimental control of a laboratory. In VR, we can modify design variables (lighting, materials, signage placement, etc.) on the moment and immediately see how those changes affect visual attention, something impossible to do in a real building. Early research shows that VR-based ET offers insights comparable to real environment studies for certain tasks (wayfinding), though we must be cautious about differences in depth perception and user interface distractions in VR.

Overall, the choice of ET system should correspond to research goals. If the priority is ecological validity and understanding natural behavior, mobile or VR systems are better despite their data noise. If the aim is detailed analysis of visual preferences or comparisons of design details, steady desktop trackers may be more appropriate.

Recognizing these compensations is part of the current methodological discourse in ET studies. Crucially, as technology improves, we anticipate that many of these limitations will diminish. Next-generation eye trackers promise higher accuracy, better outdoor performance, and multi-user capabilities, which will further solidify the role of ET in research and practice.

Challenges and Gaps

In this chapter we will provide a critical evaluation of the state of the research by examining whether some of the results of the studies previously analyzed are contradictory, or if certain topics have been neglected.

The article by Henderson and Hollingworth [34] remains a high-quality, self-critical survey that maps the theoretical field while clearly exposing empirical weak spots. Its main limitations stem from the irregular quality of the primary literature they had to review, and not from the reviewing process itself. The internal contradictions they expose are not a fault of the study but a precious indication of where the field needed, and in many cases still needs, better experiments.

Rusnak and Rabiega [28] offer an enlarged vision of how ET could enrich architectural education. The strength of the article lies in its conceptual synthesis and straightforward inventory of practical difficulties. Its weaknesses stem from the lack of systematic empirical data and several self-contradictory claims about what ET can, and cannot, deliver. It is an inspirational paper which now requires rigorous follow-up experiments.

Jam et al. [59] advance façade research by intertwining architectural theory with modern ET metrics and an ingenious layer-based coding scheme. The study is well reported, and its self-critical tone is stimulating. Nevertheless, its modest, monocultural sample, coarse preference scale, and interpretative leaps (mainly around “cognitive load”) limit the strength of its conclusions. Some findings are even contradictory: experts are claimed to make more effort yet supposedly process more efficiently, and present scan-paths that change with façade material. In general, this paper presents a valuable hypothesis, particularly about how training redirects attention from decorative to contextual cues, but this needs replication with larger and more diverse collaborators and tighter stimulus control. Finally, we recommend that the use of ET in educational contexts should first test its value to train architecture students into a more critical and expert way of looking at architecture.

Vartanian et al. [5] innovate by pairing factorial architectural manipulation with simultaneous esthetic and approach judgments inside a scanner. This study is methodologically adventurous and confirms that higher ceilings and openness look better. Yet, the sample is small, the fMRI thresholding is liberal, and the imagery is quite uncontrolled. Future studies with immersive VR, larger and diverse samples, and finer-grained rating scales are needed.

5. Guidelines for Education and Field Use of Eye-Tracking in Architecture

While recent studies have advanced the theoretical and empirical understanding of ET applications in architecture, there is a lack of operational guidance on how to effectively implement ET in educational and in situ research contexts. To address this gap, we propose two sets of practice-oriented guidelines, one focused on architectural pedagogy and the other on field studies, based on the evidence reviewed in this article.

5.1. Best Practices for Using Eye-Tracking in Architecture Education

The integration of ET into architecture education has been highlighted as an opportunity to foster reflective learning, deepen understanding of user-centered design by students, and align academic training with evidence-based practices (see

Table 1. Best practices for using eye-tracking in architecture education. (Sources: Rusnak and Rabiega [28], with methodological guidance from eye-tracking literature).

Recommended Practice	Purpose or Benefit
Embed ET exercises (e.g., building walk-throughs or peer design reviews) in coursework	Promotes reflective learning and evidence-based design by revealing where students focus attention
Highlight expert vs. novice gaze patterns in critiques	Deepens understanding of user-centered design by students by illustrating perceptual differences
Integrate ET projects and assignments into the curriculum	Broadens awareness of how designs are perceived by users by students
Plan for equipment constraints (cost, lab space)	Addresses practical implementation barriers and sets realistic project scope

) [28]. However, its pedagogical value depends not only on technological availability but also on methodological intentionality. The following practices are grounded in findings across multiple studies:

5.2. Checklist for Conducting In Situ Eye-Tracking Studies in Architecture

Conducting eye-tracking studies in real or realistic architectural contexts involves numerous technical and ethical considerations. Based on challenges identified by Mahmoud et al. [40], Zhipeng and Pesarakli [45], and Holmqvist and Andersson [2], we propose

Table 2. Checklist for conducting in situ eye-tracking studies in architecture. (Sources: Rusnak and Rabiega [28], with methodological guidance from eye-tracking literature).

Step/Item	Purpose or Rationale	Considerations or Requirements
Define research objectives/questions	Focus the study and align methods with goals	Formulate clear research questions on visual perception or navigation in built environments
Select study design and equipment	Choose ET hardware and setting to match objectives	Balance ecological validity vs. experimental control (e.g., VR vs. real-world; screen-based vs. mobile trackers)
Establish participant criteria	Ensure a representative, consistent sample	Screen for vision or cognitive issues; obtain informed consent; consider participant fatigue and comfort
Calibrate and test equipment	Maximize data accuracy and reduce error	Perform individual calibration for each participant; monitor and correct calibration drift; check for data loss (especially outdoors)
Conduct the eye-tracking session	Collect gaze data under real-world conditions	Monitor data quality in real time; minimize head/body movements; control lighting and distractions as much as possible
Analyze gaze data	Identify attention patterns quantitatively	Compute fixation counts/durations and scan paths; exclude blinks or noise; use areas-of-interest or heatmaps as appropriate
Triangulate and interpret results	Contextualize gaze with other measures	Supplement ET data with surveys or interviews to explain visual behavior; interpret findings in the architectural context

to support robust data collection and analysis:

These operational guidelines aim to enhance methodological transparency, data quality, and pedagogical relevance in future eye-tracking studies. As ET technology becomes more accessible and embedded in academic and professional settings, these best practices can contribute to a more informed, reflective, and user-centered architectural culture.

6. Conclusions

Integrating ET into research and practice offers a transformative method to merge design intuition with scientific evidence. This study has highlighted how ET can enrich our understanding of the human visual experience of architecture. By objectively capturing where people look, for how long, and in what sequence, ET provides insight into the minds of the users as they encounter architectonic contexts. Architects and researchers can utilize this to ensure that the intended focal points of a design catch the gaze and that critical information (wayfinding cues or safety features) is not being overlooked.

In essence, ET data serves as feedback on the efficiency of design elements. For instance, if a striking lobby artwork was meant to impress visitors, but ET shows most people walking past it without a glance, designers may reconsider its placement or emphasis. Likewise, if students do not notice certain features that experts consider important, educators can use that insight to adjust how design principles are taught. The comprehensive value of ET in architecture lies in its capacity to validate and inform design decisions with empirical user data, promoting a more evidence-based and user-centered design process.

In practical terms, our findings lead to several actionable recommendations for different stakeholders in the field. For educators, we recommend incorporating ET exercises and projects in architecture curricula. Even simple experiments, such as having students use an eye-tracker while exploring a building or reviewing designs of peers, can reveal biases in observation and prompt discussions on why certain elements attract attention. This practical approach can cultivate more mindful observers. As students see the discrepancies between what they and other users notice, they learn to design with the gaze of the user in mind. Rusnak and Rabiega [28] note that such integration not only broadens knowledge of perception but also encourages the interest of the students in the research dimension of design.

For practitioners (architects and urban designers), we suggest adopting ET as a tool in the design evaluation and refinement process. Before finalizing a design, conducting ET studies with representative users can generate intuitions that traditional critiques or client feedback might overlook. Practitioners could use mobile ET in a full-scale mock-up or VR walkthrough of a new space to check if people fail to notice an important sign or consistently get visually drawn to an unintended area. With the increasing availability of portable and user-friendly ET devices, even small firms can consider collaborating with specialists or universities to perform such user tests. The data gathered can inform interactive design changes, leading to architectonic contexts that are not just artistically compelling but also intuitively navigable and engaging.

For researchers, the continued development of ET in architecture should focus on both expanding the scope of studies and addressing current limitations. There is a need for more cross-disciplinary collaboration, working with vision scientists to interpret complex gaze patterns, or with data experts to handle the big data aspects of ET (a single study can collect thousands of data points per minute).

Future research should also prioritize triangulation of methods. By combining ET with other techniques like biometric sensors (to measure stress or arousal), brain imaging, or post-experiment interviews, researchers can create a richer picture of human–building interaction, linking where people look to how they feel or decide. Such multimodal approaches will help overcome the interpretation ambiguity of only using gaze data. Moreover, replicating and extending studies to diverse contexts and populations is important. As Aalto and Steinert [4] recommend, we should undertake systematic replication studies on key phenomena, verifying if, in different cultural contexts, architects consistently scan spaces differently than non-architects, or on whether certain design features universally attract attention.

These efforts will strengthen the generalization of ET findings and integrate them into foundational theory (e.g., confirming whether “universal” design principles exist in terms of human visual response).

Technologically, a clear direction for future research would be enhancing the accuracy and robustness of ET in architectonic contexts. This includes improving calibration algorithms, developing trackers that function reliably outdoors and in motion (possibly through sensor fusion or computer vision techniques), and enabling the study of multi-user architectonic contexts. Advances might allow tracking the gaze of a group of people simultaneously as they interact in an architectonic context, opening new possibilities to study social and collaborative dynamics in contexts like classrooms or public squares.

Another promising possibility is a deeper integration of VR and augmented reality (AR) with ET. As our review mentioned, VR provides a controlled yet immersive platform for testing design alternatives. Future research can further enhance this advantage by ensuring that VR-generated insights translate to real-world outcomes by using ET in VR, not just for observation, but as an interactive input (e.g., architecture that adapts in real-time to where users look, creating adaptive architectonic contexts).

Lastly, the field would benefit from developing open databases of ET data and standardized protocols, so that different studies can be more easily compared and collectively extracted to establish broader patterns.

In conclusion, ET is catalyzing a shift in architecture towards more empirically grounded design and teaching practices. It highlights the perspective of the user in a way that was previously difficult to achieve, quantifying the subtle interaction between attention, perception, and design features. As ET hardware and analysis techniques continue to improve, we anticipate that its adoption will become more mainstream, from classrooms where students improve their designs based on viewer gaze data, to professional design firms validating concepts through “eye-tracked” user experience testing.

Finally, the integration of ET in architecture enriches the ability of the discipline to create architectonic contexts that are not only esthetically pleasing and functionally efficient, but also profoundly aligned with innate patterns of human vision and cognition. This synergy between technology and design research announces a future in which architectonic contexts are developed with a deeper understanding of their users, leading to a more intuitive, comfortable, and human-centered architecture.

To conclude, we think it is pertinent to connect this investigation with the work and investigation of Jan Gehl. Gehl marked an important shift in architecture, from a metalanguage to a user centered design. This shift was (and is) crucial to architecture and we wish to keep walking this path. The convergence of contemporary eye-tracking research with the human-centric design principles of Jan Gehl underscores a shared commitment to architecture at the pedestrian scale. As this review has shown, ET has matured into an evidence-based lens for creating intuitive, legible, and human-centered environments. This scientific approach resonates with the long-standing emphasis on human-scale design and sensory experience in public spaces of Gehl [66]. Both perspectives prioritize understanding how people perceive and navigate the built environment. ET studies reveal where occupants focus, for how long, and in what sequence, providing objective insights that complement designers’ intuition. Likewise, Gehl begins with the “spaces between buildings,” observing how pedestrians move and feel at eye level and walking speed [66,67]. In effect, the technological precision of eye-tracking amplifies the qualitative observations by Gehl: both confirm that architecture must speak to our senses and scale of movement to be truly responsive and comfortable for its users. The result is a unified vision of design—rooted in data and human experience alike—that advocates lively, people-oriented places rather than abstract edifices.

Crucially, eye-tracking now offers empirical support for many of the theories on human comfort, engagement, and social interaction in cities from Gehl. Gehl has posited that the quality of an environment directly influences how long people linger and whether optional, recreational activities blossom into social encounters [68]. Eye-tracking metrics help validate these insights by objectively indicating which design features captivate attention and invite prolonged engagement. For example, Gehl observed that rich, 5 km/h details at street level (varied façades, doors, and signage) enrich the pedestrian experience [67]; ET studies corroborate this by showing that human-scaled, detail-oriented frontages consistently draw and hold the gaze of the viewers, whereas monotonous “60 km/h” streetscapes fail to do so [67]. Moreover, by capturing where people naturally look in a plaza or streetscape, ET can confirm that “comfort features”—such as seating, greenery, or clear wayfinding cues—are noticed and utilized as intended. In turn, this feedback allows architects and urban designers to refine spaces so that critical information is not overlooked and pedestrians feel at ease and oriented. The synergy between the humanistic framework of Gehl and ET analytics thus heralds a more evidence-based approach to human-scale urban design. With emerging methods even enabling multi-user gaze studies in public squares, designers can quantitatively examine the “life between buildings” that Gehl sees as the heart of city life [68]. In sum, integrating eye-tracking into architectural research and practice reinforces the vision of Gehl with scientific rigor—ensuring that our cities are not only designed for people, but are empirically tested and continuously improved to maximize comfort, engagement, and vibrant social interaction.