Based on the background presented so far, this scoping review formulates the following research questions: What is the applicability of pupillometry in the study of lexical access? What is the usefulness of this neurophysiological measure, and what are the specific aims of this area of research? What psycholinguistic methods, tasks, and procedures are used? Are the results of lexical-pupillometric research sensitive, reliable and replicable? What are the projections and limitations of this research area?

Our approach to answering the research questions involved creating a protocol (Moher et al., 2015) that aligns with the objectives of a scoping review. This type of review aims to determine the scope and methodology of the literature on a specific topic and then provide a comprehensive summary of the available evidence (Munn et al., 2018). We developed eligibility criteria considering the research context and used critical concepts for the initial search phase. Finally, although the amount of research on lexical processing and pupillometry appears to be scarce, no grey literature was included in this review.

This scoping review was designed, conducted, and reported following the PRISMA Extension for Scoping Reviews (PRISMA-ScR) checklist and explanation (Tricco et al., 2018), and was supervised by one of the researchers (GL). The process included four stages: (1) search strategy definition, (2) literature screening, (3) data extraction, and (4) data synthesis and analysis.

Five of the 16 articles included in this review assessed the applicability and usefulness of pupillometry in visual word recognition (Table 2). In particular, 1 study (Haro et al., 2017) aimed to determine whether pupillary response is sensitive to lexical frequency and whether this effect is related to lexical access. Moreover, 2 articles focused on pupillary behavior during word recognition in bilinguals (Guasch et al., 2017; Toivo and Scheepers, 2019). Finally, 2 additional articles examined pupillary dilation (cognitive effort) in different types of readers during word recognition and subsequent reading (Shechter and Share, 2021; Shechter et al., 2022).

In the first study, Haro et al. (2017) assessed pupillary response in a traditional lexical decision task (LDT, experiment 1) and a delayed lexical decision task (LDT delayed, experiment 2). During the experiments, 60 participants (30 per group) had to discriminate between pseudowords and words while their pupillary response was monitored. The words had different lexical frequencies (low, medium, or high). The results of the study suggest that pupillary response is a sensitive indicator of word frequency, with low-frequency words causing significantly grater pupil dilation than high-frequency words, regardless of the LDT type. The authors argue that pupil dilation would represent a genuine effect of lexical processing and could be considered an objective measure for assessing word recognition, as it would not be mediated by response mechanisms accessory to LDT (i.e., executive control, inhibition or sensorimotor aspects).

Among the articles focusing on bilingualism, Guasch et al. (2017) studied how bilinguals’ pupillary response is affected by words that fulfill the cognate condition, i.e., words from different languages with similar spelling and identical meaning. Using LDT, pupil dilation was recorded in 35 bilingual native speakers (Spanish-Catalan) when they processed words in the identical cognate/non-identical cognate/non-cognate condition. The study showed that the non-cognate word condition (words with no spelling overlap but identical meaning) produced a significantly greater pupil dilation than the other experimental conditions. The research provided evidence that pupillometry is sensitive to specific word characteristics (semantic-orthographic congruence/incongruence) in bilingual native speakers. Along the same lines, Toivo and Scheepers (2019) assessed the sensitivity of the pupillary responses to words of high emotional arousal words in L1 and L2 using LDT. They included 96 monolingual and bilingual young adult participants. The study revealed that emotionally loaded words in the native language (L1) produced a stronger pupillary response (increased dilation) than emotionally loaded words in L2, possibly due to the richer semantic representation created by emotional words in L1. According to the authors, pupillometry could be an objective and valuable tool for assessing the recognition of emotionally loaded words.

Finally, 2 articles were compiled that examined the applicability and usefulness of using pupillometry as an indicator of cognitive effort in word recognition in diverse type of readers. In example, Shechter and Share (2021) examined the pupillary response during a task with known words and pseudowords (letter strings of different lengths) in 48 university students and 48 schoolchildren. Consequently, longer pseudowords resulted in greater pupillary dilation, indicating increased cognitive effort of both experienced readers (university students) and novice readers (schoolchildren), both for oral and silent reading task. The authors project that pupillometry may offer a more sensitive moment-by-moment glimpse into the dynamics of word recognition (including developmental, interindividual, and intraindividual variation). In a second study, Shechter et al. (2022) assessed the sensitivity of the pupillary response during recognition and reading aloud of a series of words in children (N = 34) and young adults (N = 34), who were further classified into fast/slow readers. The results showed that children showed more significant cognitive effort (as evidenced by a greater pupillary response) than adults during word recognition and reading aloud. However, no differences in cognitive effort were observed between fast and slow readers in any age group. According to the authors, the pupillometric index would offer objective hints about how reading ability develops starting in the earliest phases of life.

Seven of the 16 articles reviewed (Table 3) evaluated the applicability and usefulness of pupillometry in activation and semantic word-processing tasks. One article relied on embodiment theory to show whether there is a significant pupillary response when processing words whose meaning is associated with a higher or lower level of brightness (Mathôt et al., 2017). Five additional articles examined the behavior of the pupillary response when establishing congruence or association judgments between pairs of words with specific semantic and orthographic features (Geller et al., 2019; Egan et al., 2020; Reilly et al., 2020; Egan et al., 2023; Haro et al., 2023). Another study (1) examined pupillary response sensitivity to determine whether the semantic priming effect is present in children under 24 months of age (Angulo-Chavira and Arias-Trejo, 2021).

In the first study reviewed, Mathôt et al. (2017) assessed pupillary response in a semantic recognition task within the framework of embodied language theory. In the experiment, participants (30 in total) were tasked with identifying words that were semantically linked to brightness, darkness, neutrality (words that did not evoke brightness or darkness), as well as animals (words whose meaning is not associated with any aspect of brightness or darkness). Participants were instructed to press the space bar when animal-related words appeared on the screen, while their pupillary response was monitored throughout the trials. The pupillary response, which reacts to changes in illumination, showed a smaller dilation when words linked to brightness were activated but a larger dilation when processing words related to darkness. The authors confirm that the pupillary response is consistent with embodied language theory, as semantic activation of word meaning was sufficient to elicit a response without needing an external light stimulus.

In the case of articles focusing on pupillary response to judgments of congruence or semantic relatedness between word pairs, Egan et al. (2020) used Event-Related Potential (ERP) and pupillometry to examine how alliteration (similar phonological form) between words improves attentional interactions during reading. Twenty adult participants were asked to make semantic judgments between semantically congruent/incongruent word pairs that also satisfied the alliteration/non-alliteration condition. The results showed that the pupil response was larger for word pairs that met the congruence condition (alliterative or non-alliterative). The research also found that the interaction between alliteration and congruence led to a continuous increase in pupil dilation when related words appeared in a sentence. According to the authors, alliteration, as measured by ERP and pupillometry, can influence semantic processing by supporting concept activation during comprehension errors. The same authors (Egan et al., 2023) evaluated the sensitivity of pupillometry to measure the influence of phonological relatedness on semantic congruence judgments in readers with dyslexia. 38 participants with (19) and without dyslexia (19) were asked to semantically process orthogonally manipulated word pairs for semantic congruence and alliteration (phonological form). Participants had to indicate whether the two words presented were meaning-related by pressing a button. As a result, pupil dilation was consistently lower in dyslexic readers. According to the authors, people with dyslexia show reduced engagement of attentional processes during reading, making pupillary response a sensitive measure for detecting differences in lexical processing among readers.

Geller et al. (2019) examined the utility of the pupillary response in a processing task of taxonomically or thematically semantically related word pairs, whose relatedness could be weak/strong. Sixty undergraduates had to make judgments about the relatedness/unrelatedness of the word pairs. The results showed greater pupil dilation in weakly related pairs than in strongly related pairs (particularly taxonomic pairs), where greater pupil dilation would require greater cognitive effort. The pupillary data would suggest that semantic control demands would be determined primarily by the strength of the semantic relationship (weak/strong) and not necessarily by the type of relationship (taxonomic or thematic). In a similar study, Haro et al. (2023) examined whether the pupillary response was sensitive to differences between polysemous and homonymous words when performing tasks involving different processing costs. The authors conducted four experiments on a sample of 117 adult subjects. Specifically, in Experiment 3 (n = 28), participants had to solve a semantic categorization task, and in Experiment 4 (n = 24) they had to determine whether a word had more than one meaning. Greater pupil dilation was only observed for those words that were congruent to the semantic category evaluated (Exp.3) and when ambiguous words were processed (Exp.4). No effects were observed when comparing polysemous and homonymous conditions. The authors point out that there is insufficient evidence to demonstrate differences in the processing of polysemous/homonymous words. However, the processing of ambiguous words and those that are of the same semantic category, would generate greater cognitive cost and therefore greater pupillary response.

In the same line, but with different aims, Reilly et al. (2020) employed pupillometry to assess the impact of transcranial stimulation on the processing of taboo and non-taboo words. Adult participants were required to read aloud words of taboo/non-taboo meaning before and after transcranial stimulation. As a result, transcranial stimulation on the right cerebral hemisphere impacts taboo word processing, which is reflected in increased pupillary dilation. In their conclusion, the authors state that the pupillary response to taboo words after transcranial stimulation results from the activation of emotional valence and the person’s unconscious inhibitory control when processing socially inappropriate words. Finally, Angulo-Chavira and Arias-Trejo (2021) explored the pupillary response when executing a semantically mediated priming task within the same level (i.e., cat [prime] - mouse [mediator] - cheese [target]) to determine whether this semantic effect is present in children younger than 24 months. Twenty-seven children observed eight experimental trials set up on a computer. The trials were divided into two conditions: the related condition (target and cue words had an indirect relation) and the unrelated condition (target and cue words had no semantic or associative relation). The pupillary response shows that children present mediated priming effects from 24 months onwards. However, the authors argue that there may be inhibitory mechanisms that limit their activation from an early age.

Four of the 16 articles found (Table 4) evaluated the applicability and usefulness of pupillometry during word retrieval. Three articles address pupillary response in verbal fluency tasks (multiple word retrieval) in various populations, and the remaining one addresses pupillary behavior in the presence of the tip-of-the-tongue states (TOTs) in adults.

El Haj et al. (2020) assessed pupillary response in a phonemic verbal fluency task (retrieval of multiple lexical items). Forty-five young adult participants were asked to retrieve all words beginning with the phoneme “p” and then to count numbers aloud while their pupillary dilation was tracked. As a result, greater pupillary dilation was obtained for the verbal fluency task compared to the counting task. The authors suggest that the increased pupil dilation during verbal fluency can be attributed to the cognitive requirement of the task, as participants were required to retrieve words, suppress irrelevant information, and explore alternative options. The same research team (El Haj et al., 2022) repeated the experiment with a group of older adults (n = 45). The results indicated that pupillary dilation increases among older adults during the verbal fluency task. For both studies, the researchers project that pupillometry could be helpful as a method of ecologically assessing lexical retrieval skills at different stages of the life cycle. In a recent publication, El Haj et al. (2024) evaluated individuals diagnosed with bvFTD compared to controls. Consequently, individuals with bvFTD showed less pupil dilation during the letter and category verbal fluency task (animals) compared to the control group. However, when comparing both tasks with counting numbers, the bvFTD group showed a greater pupillary response. The authors explore pupillometry’s applicability and ecological importance in investigating language processing in persons with dementia.

Finally, Ryals et al. (2021) explored how sensitive the pupillary response is during TOTs. They evaluated 46 undergraduates who were asked to answer pre-made questions. Participants had to report the sensation of TOTs when they could not retrieve the words associated with the requested questions. The results showed that TOTs generated a significant pupillary response (increase in diameter). The authors suggest that the increase in pupil size supports the notion that TOTs are linked to feelings of curiosity and the desire for knowledge, with the degree of pupil dilation corresponding to the amount of information retrieved. Therefore, pupillometry would be an appropriate and valuable tool for studying these linguistic phenomena.

The general interest in cognitive neuroscience in studying systematic changes in pupillary dilation is based on the fact that it is a noninvasive, objective and automatic neurophysiological measure (Hepach, 2023), which can reflect-if well controlled-the behavior of multiple cognitive constructs in diverse populations and life cycle stages (Laeng et al., 2012; Kafkas and Montaldi, 2015). In this context, the results of this review support that the pupillary response evoked by language processing tasks-and reflecting neuronal activity-has become an emerging area of development in experimental psycholinguistics over the last decade. However, its applicability is not exclusive to language processing. Our findings suggest that around 80% of the articles recovered during the identification phase were associated with various cognitive domains such as memory, attention, motivation, and emotions, where language tasks were merely one aspect of their data-gathering strategies and did not play a significant role. Out of the total, only 20% of the studies explored language processing, covering various levels (syntax, auditory recognition, speech perception, and lexical access) across different age groups and in neurotypical individuals or with neuro-cognitive disorders (congenital and acquired).

In particular, the applicability and usefulness of pupillometry as a method to assess cognitive processes have been demonstrated in several research fields (Ariel and Castel, 2014; Gomes et al., 2015; Kafkas and Montaldi, 2015; Larsen and Waters, 2018; de Winter et al., 2021). These areas comprise surprise and expectancy paradigms (Pätzold and Liszkowski, 2019), memory and attention tasks (Larsen and Waters, 2018; de Winter et al., 2021), or arousal and motivational states (Hepach et al., 2019). Regarding language and lexical access, we found that only 16 articles fully met the established criteria. Nonetheless, the careful examination of their methods, pupillometric analysis, statistical procedures, and outcomes provides sufficient evidence to assert that pupillometry is a versatile, valuable, and sensitive method for assessing linguistic processes involved in word recognition, retrieval, and semantic activation, and that it can be applied to any population, age group, and native or bilingual speakers. Indeed, the studies demonstrated that pupillary changes – measured mainly using peaks and dilation latencies – effectively reflect lexical activity, which allowed effects to be tested in 94% of the reports analyzed.

Furthermore, the present review demonstrates that the research purposes (objectives) in lexical access and pupillometry are also very diverse and varied, ranging from pupillary response in emotional word recognition in bilinguals (Toivo and Scheepers, 2019) to pupillometric behavior in lexical fluency tasks in individuals with bvFTD (El Haj et al., 2024). The same applies to the designs, techniques, and procedures applied, where pupillometry can be effectively used for word retrieval and semantic judgment tasks, as well as LDT (or its variations). In addition, there are differences in the cleaning, analysis, and interpretation of the pupil data. These variations include using different analysis windows, the basal pupil diameter used, the pupil variable examined (maximum pupil dilation vs. maximum latency), and other factors. Thus, the emerging development and the diversity of methods observed in this field, which can be considered an advantage, do not yet allow definitive and solid conclusions to be drawn in this area.

On the other hand, although the collected background information demonstrates the usefulness of pupillometry in addressing various research problems in lexical access, it would be incorrect to assume that any pupillary response indicates lexical activation. From a methodological point of view, it is always necessary to “isolate” other cognitive factors associated with pupillary dilation, such as surprise, attention, or task-associated motivation (Hepach, 2023), as they could mask the expected pupillary response. Hence, while pupillometry is undeniably useful for studying lexical access and shows promising growth potential, it still requires improved precision and standardized conditions of use that demonstrate that it truly represents the pupillary response obtained, as we are not entirely sure whether it is an exclusive response to the cognitive effort expended in the task (Van der Wel and Van Steenbergen, 2018), or a response to the attentional component present in the cognitive function performed (Strauch et al., 2022), or simply a specific response for language processing.”

In sum, this review validates the applicability of pupillometry as a sensitive and valuable technique for assessing various lexical levels, which is a positive attribute of this neurophysiological measure. However, due to the varied purposes, objectives, and data analysis approaches of the reviewed studies, we cannot establish definitive conclusions in this area. Multiple reasons could explain this: (1) Pupillometry is a recently popularized and recognized technique; (2) pupillary dilation is a very volatile signal whose control and instrumental measurement have recently achieved better management (Hepach, 2023); (3) the interest in its application and use is limited to specific research groups rather than large collaborative networks, who turned to pupillary dilation measurement as a practical and conceptual solution to their research problem (Schmidtke, 2018). In this sense, defining more standardized purposes for applying pupillometry in lexical access would enable us to produce far more accurate research purposes, enhancing the vigor and robustness of this study area.

There is a consensus that most studies in experimental psycholinguistics use small sample sizes, which contrasts with the large number of experimental trials used to favor greater statistical power (Perea and Rosa, 1999). In addition, these studies often include young, normotypical participants selected for interest, with the aim of testing hypotheses in a ‘healthy’ brain (Rojas et al., 2022). In this respect, lexical-pupillometric studies are no exception. For example, independent of the level of processing assessed, all articles exhibited bounded sample sizes in each experiment performed (i.e., Recognition: n = 35, Guasch et al., 2017; n = 96, Toivo and Scheepers, 2019; n = 68, Shechter et al., 2022. Semantic activation: n = 20, Egan et al., 2020; n = 30, Mathôt et al., 2017; n = 60, Geller et al., 2019. Retrieval: n = 45, El Haj et al., 2020, 2022; n = 46, Ryals et al., 2021), but they used a large number of experimental trials. Moreover, like the “more traditional” studies, the evaluated populations were preferably young university students with normal cognition, vision, and hearing. This situation varied when the object of study involved specific age groups or clinical conditions (n = 19 persons with dyslexia, Egan et al., 2023; n = 27 children under 24 months, Angulo-Chavira and Arias-Trejo, 2021; n = 24 patients with bvFTD, El Haj et al., 2024). Given the benefits of pupillometry (i.e., a non-invasive, objective, and free of conscious responses), sample sizes and types of participants observed in the studies reviewed, it is possible to assume that this technique could be applied in any population, particularly to individuals with complex motor, behavioral, or cognitive control, such as very young children (Hepach et al., 2019), older people and individuals with neurodegenerative diseases (El Haj et al., 2022, 2024). This fact reveals another advantage and endorses the potential of this neurophysiological tool.

Regarding the experimental tasks used in lexical-pupillometric studies, at the level of word recognition and similar to traditional studies in the field, two articles used classical LDTs (Guasch et al., 2017; Haro et al., 2017), and two additional studies used word recognition and subsequent reading methods (Shechter and Share, 2021; Shechter et al., 2022). On the other hand, regarding the activation and semantic processing level, only semantic judgment tasks were stated in several articles (Geller et al., 2019; Egan et al., 2020; Reilly et al., 2020). At the retrieval level, verbal fluency (El Haj et al., 2020, 2022, 2024) and naming by definitions (Ryals et al., 2021) tasks were preferentially used. Therefore, the assessment paradigms in the reviewed lexical-pupillometric studies are the same as the traditional ones. Additionally, they share similarities in the number of items used, the organization of each trial, the number of blocks, the stimulus presentation latencies, and the recorded behavioral responses (RT and accuracy). Similarly, lexical variables like frequency (high/low), lexicality (words/pseudowords), syllable length, polysemy, phonological form, and prime type, among others, are also subject to control. In this sense, pupillometry’s easy adaption to conventional research paradigms in experimental psycholinguistics supports its numerous uses and provides further evidence for its effectiveness in examining lexical access.

The situation becomes more confusing when analyzing and comparing the methods used to collect, control, and analyze the resulting eye data. Although, it should be recognized that basic points are consistent across articles: (1) pupillary data were recorded before, during, and after the presentation of the target stimulus; (2) areas of interest were previously identified according to the purpose of each examination, and (3) two classes of pupil measurements were preferably calculated: maximum dilation and maximum pupil latency (this is consistent with the language processing and pupillometry studies referenced by Schmidtke, 2018); it should also be noted that there are other aspects that need to be described and addressed in more detail (see point 4.4. Recommendations). In conclusion, although the necessary details for pupillometric control and analysis are reported, there are also some methodological omissions, probably made but not explicitly stated, which are necessary to replicate the studies presented here.

Hepach’s work (Hepach, 2023) defines some requirements for more effective pupillary dilation control at the experimental level. In this sense, the pupil response would not only reflect a particular cognitive response, but would also be the result of: (1) light adaptation to the environment and the focused visual field, (2) eye and head movements (Brisson et al., 2013), (3) changes in tonic arousal state (cardiac or cutaneous arousal state; Anderson and Colombo, 2009), and (4) the cognitive state of the participants (Eckstein et al., 2017). In this context, only a small number of the reviewed articles offer partial details regarding the light control of the environment, the particular visual field of focus, the participants’ head posture, and their state of arousal; except for the cognitive state, which was always stated and controlled. The lack of control of these experimental conditions and, in particular, the light verification during the application of the experiments (Schmidtke, 2018) are probably some of the main weaknesses observed in the reviewed studies.

As mentioned, determining a priori that pupil size increase reflects only lexical activity (or word processing) may be risky and highly inaccurate. In contrast, pupillary responses may simultaneously encompass other cognitive processes (i.e., cognitive load, surprise, motivation; Larsen and Waters, 2018; Hepach et al., 2019; Pätzold and Liszkowski, 2019) and depend on external elements such as ambient illumination and intrinsic characteristics to the participant (Brisson et al., 2013; Eckstein et al., 2017). Therefore, the sensitivity and reliability of pupillometry will depend to a large extent on methodologically rigorous and specific experiments and their subsequent analysis. In this regard, all the articles reviewed proved to be very strict both in the approach to the lexical task and in the subsequent pupillometric analysis (although, as already described, in some cases, specific methodological parameters necessary for replication were not explicitly stated). Therefore, apart from some observed weaknesses, the analysis allows us to argue that the pupillometric results obtained seem to represent, in whole or in part, the lexical processing of words.

At the statistical level, the analyzed lexical-pupillometric results (i.e., maximum pupil dilation or maximum pupil latency) are quite clear and significant at the different levels tested. For example, in word recognition tasks, statistically significant (p < 0.05) pupillary dilations were evidenced when recognizing low-frequency words (Haro et al., 2017), words without cognate status (Guasch et al., 2017), emotional words in L1 (Toivo and Scheepers, 2019) and when processing pseudowords (Shechter and Share, 2021). Concerning semantic activation, significant pupillary dilation has been observed in the processing of words with low semantic relatedness (Geller et al., 2019), in the activation of taboo words (post-transcranial stimulation, Reilly et al., 2020), and in the processing of words that are semantically related or consistent to a semantic field (Egan et al., 2020; Haro et al., 2023). At retrieval level, significant pupillary dilation (p < 0.05) occurs when performing verbal fluency tasks and when TOTs are present (El Haj et al., 2020, 2022, 2024; Ryals et al., 2021). In sum, although the number of articles reviewed is relatively small to determine exactly how suitable pupillometry is for the study of word processing, it is striking that 94% of the articles reviewed showed the expected effects, which seems to support the idea that the technique does indeed appear to be sensitive enough for the study of lexical processing.

However, the same cannot be said for the reliability of pupillometric measures. With the evidence reviewed in the present review, it cannot yet be determined whether pupil dilation is specifically and uniquely related to language processing. It is also uncertain whether the pupil dilation levels obtained in the applied tasks are not at least partially influenced by other cognitive processes, such as motivation, arousal, and attention, as evidence from other fields shows (Hepach et al., 2019; Pätzold and Liszkowski, 2019). In this sense, the influence of other cognitive processes on pupillometric measures is not in itself a problem; many of the linguistic processes are not exclusively linguistic and a large number of factors are involved in both the comprehension and production of language. However, it is necessary to determine the relative weight of each factor in the pupillometric measures obtained in order to better isolate specific aspects of language processing.

In this context, we hypothesize that the different neuromodulatory systems responsible for pupil activity (PON/SC/LC, Joshi and Gold, 2020; Joshi, 2021) operate as interconnected systems that generate specific pupil responses in the presence of different cognitive functions (i.e., attention, language, emotion, working memory). Therefore, we think word processing (or lexical access) may generate pupil dilation patterns specific to this function (i.e., reflected in specific temporal dilation window, mean dilation diameters or particular dilation peaks). Thus, it is likely that there are distinct dilation components for linguistic tasks related to word recognition, lexical retrieval, and semantic activation. However, these specific pupil dilation components need to be isolated and distinguished from other response components (similar to the behavior of Evoked response potential/ERP components), such as cognitive effort, surprise, attention, or memory, which are likely to manifest themselves through pupil response patterns that are different and distinct from the ‘linguistic’ pupil response. Thus, determining which cognitive-neural processes are actually represented by changes in pupil size has become one of the major current challenges for cognitive neuroscience.

One way to ensure the reliability of lexical-pupillometric results is for studies using this measure to generate their own experimental designs. In other words, these experimental designs should be intended as pupillometric studies from the initial stage of the experimental design and not at later stages (Hepach, 2023). Hepach (2023) mentions that it would not be advisable to ‘salvage’ data from a completed (and for other purposes) eye movement study to perform a supplementary or secondary pupillometric analysis, as there are numerous artifacts and experimental manipulations that could alter the results. Consequently, any study examining pupillary changes during word retrieval, semantic activation, or recognition should be designed from the beginning as a pupillometric study. This includes considerations for sample size, basal pupil diameter definition, analysis window duration, control over blink frequency, saccades, and other artifacts, control over illumination (i.e., matching the luminance of the stimuli or using the same image in all experimental conditions), excitability and cognitive characteristics of the persons assessed, among other aspects.

The results of the present review allow us to propose some basic recommendations for using pupillometry in lexical access tasks. First of all, considering the multiple applications of this neurophysiological measure, its different forms of measurement and the different alternatives in data analysis, we suggest that, before formulating the experiments, the researcher should: (1) deepen his or her physiological knowledge of the pupillary response; (2) inquire about technical and procedural aspects already established in lexical-pupillometric experiments; and (3) formulate or replicate objectives based on previous studies in the field, which will allow improving the consistency and stability of research purpose in different populations and contexts.

Secondly, we have found certain methodological and procedural weaknesses in several lexical-pupillometric studies that should be described and addressed in more detail. For example, more precise control over the number of blinks, saccadic movements, and other artifacts is recommended. It is known that data ‘contaminated’ by blinks (N > 5) and saccades (N > 5) in five or more trials throughout the experiment can lead to erroneous results (Kafkas and Montaldi, 2015; Carbajal et al., 2018; McKee et al., 2018). Similarly, the differences in measurement units used for pupil dilation (arbitrary chamber units vs. metric units) and the lack of explanation for choosing one measurement over another must be addressed. It is also recommended to comprehensively describe the methodology used to calculate the mean number of eye fixations and the mean fixation duration for each trial. In addition, it is recommended to provide further explanation on how the basal pupil diameter was calculated: by averaging the anterior pupil diameter during the target appearance or while the fixation cross was displayed? In this sense, for future research in this area, we propose to review the methodological considerations for using pupillometry in the study of language processing, according to Schmidtke (2018).

Thirdly, there is an urgent need to replicate already published studies, increase sample sizes and standardize specific parameters of pupillary analysis in order to increase the reliability of this neurophysiological measurement (Hepach, 2023). Thus, greater and better replicability of studies and standardization of the forms of recording, administration and analysis of pupillometric results will favor the knowledge and reliability of the procedure and allow access to more consistent language processing data.

Finally, as general recommendations, Mathôt and Vilotijević (2023) proposed a practical guide for the general design of cognitive experiments using pupillometry. In their paper, they propose six basic principles for experimental design: (1) stimuli used in experiments should be constant across conditions; (2) eye position should also be constant across conditions; (3) tests should be presented at a slow presentation rate; (4) pupil size should be assessed while participants are doing nothing; (5) ambient lighting should be intermediate and adapted to the brightness of the screen; and (6) data should be stored in a single file per participant (for further details see Mathôt and Vilotijević, 2023). In addition, for the correct selection of participants, recording characteristics, stimulus and timing parameters, data accessory noise, measurement procedures, and presentation of results and plots, we suggest reviewing the paper by Steinhauer et al. (2022), “Publication guidelines and recommendations for pupillary measurement in psychophysiological studies.” Finally, we recommend reviewing the paper by Kret and Sjak-Shie (2019), “Preprocessing pupil size data: Guidelines and code,” for guidelines and technical suggestions for correct pupil data processing and statistical analysis.

The projections of pupillometry in the study of lexical access are invaluable in developing a better understanding of how people process information. First, it is an objective, non-invasive neurophysiological tool without voluntary response mechanisms (Haro et al., 2017), which provides information about the time course of the process being assessed (Guasch et al., 2017). Second, its high adaptability and usefulness are favorable for studying word recognition in different age groups, both in the native language and in L2 for bilinguals (Toivo and Scheepers, 2019), and it can be coupled with LDT or recognition-reading tasks. In addition, it allows working on semantic activation tasks, such as word retrieval, in neurotypical individuals or those with complex motor, cognitive, and behavioral control (Angulo-Chavira and Arias-Trejo, 2021; El Haj et al., 2024). Third, while specific methodological issues with pupillometric data analysis can be fixed, the reviewed studies demonstrate a high degree of rigor and thoroughness in their procedures, leading to results that seem sensitive.

Furthermore, the technology used to record and measure pupillary behavior is expected to advance in accuracy and dependability, given its volatility (Hepach, 2023). Enhancing these parameters will make it possible to record responses in greater detail, making it easier to interpret the anticipated effects. Moreover, a more systematic integration of lexical and pupillometry research with other neuroimaging technologies is necessary. Combining various measurement modalities can yield a more thorough understanding of the linguistic-cognitive processes under examination. This would not only help to increase knowledge in a normotypical population, but could also strengthen clinical research into the understanding and diagnosis of neuropsychiatric disorders, as well as the development of future biomarkers for the early detection of Alzheimer’s, Parkinson’s, autism spectrum disorders and other neuropsychiatric disorders.

Pupillometry also has limitations that have partially impeded its advancement in cognitive neuroscience. The first and perhaps most common reason is that several factors affect pupil dilation, such as ambient lighting, age, fatigue, and the person’s general state of alertness, making it an extremely volatile measure that is difficult to capture and interpret. This limitation is compounded by non-cognitive autonomic physiological perturbations (stress response or direct light stimulation of the experimental material), where distinguishing between cognitive and non-cognitive responses is challenging. An additional constraint pertains to the inter-individual variability in pupillary response (basal pupil diameter) among participants. This variability must be considered when analyzing the data since a typical response for one participant might not be typical for another.

Finally, on a technical level, pupillometry requires a thorough knowledge and handling of the software and hardware used, which is essential for reliable measurements. Inappropriate pupil calibration or reading may result in errors in the interpretation of results. A reliable (and therefore reproducible) pupillary response requires strict and controlled experimental and methodological conditions. A “light” or loose experimental design may influence pupillary response or reflect other cognitive processes. Thus, despite some limitations, pupillometry is a valuable tool in cognitive research and, in this particular case, in the study of lexical processing. By combining pupillometry results with those obtained from other measures and techniques, researchers can gain a more complete understanding of the cognitive processes under study. Of course, it is important to recognize and address these limitations in order to generate and interpret these data reliably and accurately.