Qualitative and Quantitative Measures of Joint Attention Development in the First Year of Life: A Scoping Review

Abstract

Joint attention (JA) is the purposeful coordination of an individual’s focus of attention with that of another and begins to develop within the first year of life. Delayed, or atypically developing, JA is an early behavioral sign of many developmental disabilities and so assessing JA in infancy can improve our understanding of trajectories of typical and atypical development. This scoping review identified the most common methods for assessing JA in the first year of life. Methods of JA were divided into qualitative and quantitative categories. Out of an identified 13,898 articles, 106 were selected after a robust search of four databases. Frequent methods used were eye tracking, electroencephalography (EEG), behavioral coding and the Early Social Communication Scale (ECSC). These methods were used to assess JA in typically and atypically developing infants in the first year of life. This study provides a comprehensive review of the past and current state of measurement of JA in the literature, the strengths and limitations of the measures used, and the next steps to consider for researchers interested in investigating JA to strengthen this field going forwards.

1. Introduction

Joint attention (JA) is a uniquely human skill and can broadly be defined as the purposeful coordination of an individual’s focus of attention with that of another individual in order to obtain shared attention towards a common event or object (Baldwin, 1995). JA involves two individuals coordinating mutual engagement and focus on a third-party entity (Tomasello, 1995). It is a vital skill for social competence at all ages in life and impaired JA can detrimentally affect social communication and interpersonal development.

It has been suggested that JA is the result of two complementary and interacting attention systems: (1) the perceptual attention and posterior orienting system; (2) the anterior attention system (Rueda et al., 2004). The first system is thought to enable RJA (responding to JA), while the latter enables IJA (initiating JA). The posterior orienting system, enabling RJA, is thought to develop in the first few months of life and is an involuntary mechanism that prioritizes stimuli that are meaningful. This system is supported by the parietal and superior temporal cortices which play a role in the perception of others’ eye and head orientation in addition to spatial relationships between individuals and their environments (Mundy & Newell, 2007). On the other hand, the anterior attention system, which enables IJA, develops later, and supports goal-directed attentional behavior. One neural network involved in this system is the dorsal frontal cortex which is responsible for intentional eye control (Mundy & Newell, 2007). These two different attentional systems must work together for efficient JA to take place. RJA is the first to develop and is a receptive form in which an individual follows the gaze or gestures of others to engage in shared attention with a common object (P. Mundy, 2021), while IJA refers to an individual’s use of attention-directing behavior towards a partner to capture their attention toward this common object (Cilia et al., 2020). Within this scoping review, the distinction between RJA and IJA will be used to contribute to the suggested advantages and disadvantages of the methods found in the literature assessing JA in the first year of life. Elicited behaviors to engage in JA can be verbal or non-verbal. Verbal JA cues are the use of language to share experiences with others, or overtly guide someone’s attention to an external target. Non-verbal JA cues include a variety of deictic features including pointing gestures and coordinated eye gaze (Paparella et al., 2011).

Investigating JA in infancy is especially interesting, not only because this phenomenon typically begins to develop over the first year of life but also because pre-linguistic infants are only able to use non-verbal JA cues such as pointing and modulating eye gaze. JA behaviors have been shown to emerge at approximately 6 months of age in children with typical development (Adamson, 2014) at which infants begin to coordinate attention with a caregiver for an object in their environment, such as a toy. This corresponds to the age at which infants begin to sit up by themselves, making it easier to manipulate objects and make eye contact with the caregiver. Once an infant reaches approximately 8–9 months old, they become more proficient in following caregiver attention and consequently initiating new JA. At this stage, the infant may begin to crawl allowing them to interact with more objects in their environment, leading to an increase in episodes of JA.

JA is related to a number of other social-cognitive constructs including Theory of Mind, emotion regulation and broader general attention (Adamson et al., 2019). When an individual seeks or follows a shared focus of attention, they are able to display that they have an understanding of another’s mental state. Measuring JA allows us to thus observe physical human behaviors that underlie unobservable mental states. JA also is thought to underlie the language acquisition process (Tomasello, 1995). In infancy, an inability to engage in a shared focus of attention renders it more difficult for the infant to engage in word-learning opportunities with a caregiver. This has a knock-on effect on social information processing which can impact individual differences in childhood measures of social competence, self-regulation, and general intelligence (Mundy et al., 2007). The emergence of JA skills early in life helps to scaffold these key developmental behaviors by allowing children to communicate efficiently with others and develop important social skills, building their capacity for relationships, relatability and information processing that will be vital later in life.

Some infants may experience delayed or atypical development of JA. Delayed, or atypically developing, JA can lead to a multitude of social-cognitive and social-emotional developmental issues including poor language outcomes, and impairments and delays in socio-emotional and theory of mind skills in comparison to their TD peers (Urqueta Alfaro et al., 2018). It is widely acknowledged that impairments in JA are one of the earliest signs of autism spectrum disorder (ASD) (Gernsbacher et al., 2008) and that JA behaviors tend to be delayed in children with Williams Syndrome (Hahn, 2016). Here, it is likely that impaired JA is a symptom of these later emerging social-cognitive and social-emotional issues. For example, impaired JA is one symptom that can be identified early in life and act as a prognostic indicator of ASD (Bruinsma et al., 2004). In addition, having impaired JA impacts other developmental domains and can thus exacerbate other symptoms associated with later emerging developmental issues like ASD, such as delayed language skills and unusual emotional reactions (Hodges, 2020). If JA capabilities are measured early by using the right techniques and are seen to be impaired or delayed, then potential interventions can be implemented to help minimize the developmental consequences associated with impaired JA. Measuring JA in the first year of life is thus important in identifying and understanding the early trajectories of both typical, and atypical, development in infancy in order to identify potential symptoms of later emerging developmental issues for which early interventions can be implemented.

JA can be measured utilizing different techniques, here we approach this by dividing the techniques into qualitative and quantitative methods. The qualitative assessments require some subjective interpretation in that a researcher is needed to determine the level of JA after real time or retrospective coding of a video recorded observational measure. On the other hand, the quantitative methods produce objective numerical variables that provide a measurement of degree of JA. While many studies assessing infant JA exist, they use variety of methods to assess it, illustrating that there is no widely agreed upon gold standard measure. In addition, the studies assessing infant JA include different ages and/or population groups making it hard to compare across papers what the most effective assessment measure is. Thus, the aim of this scoping review is to identify all the qualitative and quantitative methods used for assessing infant JA in both typical and atypically developing infants in the first year of life and identify the corresponding strengths and limitations of these methods. We aim to comprehensively present the past and current state of the field of JA measurement in infancy, and describe strengths and limitations of these assessments with regards to measuring JA. This information can consequently be used to guide future research on what types, or combinations, of methods should be employed. A scoping review is appropriate for this literature review and synthesis, and, to date, no scoping reviews have been conducted regarding this purpose.

2. Methodology

We conducted a scoping review (Arksey & O’Malley, 2005) that was in adherence with the Preferred Reporting Items for Systematic Reviews and Meta-analyses extension for Scoping Reviews (PRISMA-ScR; Tricco et al., 2018). Scoping reviews are most frequently used for reconnaissance on a topic and are therefore useful for clarifying aspects of a previously unreviewed subject field (Peters et al., 2020). They also allow for a broader overview of the topic at hand than other review methods (such as systematic reviews), allowing us to describe the existing evidence base fully and comprehensively (Munn et al., 2018). Thus, a scoping review was an appropriate approach in order to comprehensively assess and investigate the body of literature regarding JA assessment in the first year of life. There is a significant body of evidence-based research on the assessment of JA. However, to our knowledge, a comprehensive review on the most commonly used measures for JA and corresponding strengths and weaknesses of these methods has not been done.

2.1. Search strategy

2.1.1. Eligibility criteria

This scoping review used the following inclusion criteria: (1) Only studies that used an explicit measure of JA in the methodology were included (many papers exist researching general infant developmental capabilities, but we were only interested in measuring precise JA measures); (2) Only sources in English were included due to language constraints; (3) Theoretical reviews were ineligible; (4) Papers from all years were considered in order to get the most comprehensive overview of the existing body of literature; (5) Only papers that included infants in the first year of life were included (studies that reported a participant mean age of higher than 12 months of age were excluded). This was a cut-off point established in order to ensure that we were investigating how JA is typically assessed in non-verbal infants.

2.1.2. Information sources

Searches were conducted in PubMed, Scopus, Embase and PsychInfo databases. The final searches were conducted on 03/08/2022. The search strategies for each database are as follows: PubMed (“Infant”[MeSH] OR “neonat*”[tiab] OR “newborn”[tiab] OR “newborns”[tiab] OR “infan*”[tiab] OR “baby”[tiab] OR “babies”[tiab] OR “nursery”[tiab] OR “nurseries”[tiab]) AND (“attention”[MeSH] OR “joint attention”[tiab]) AND (“child development”[MeSH] OR “child develop*”[tiab] OR “childhood develop*”[tiab] OR “infant develop*”[tiab]). This search produced 1,242 results; Scopus TITLE-ABS-KEY(infant OR neonate OR newborn OR newborns OR baby OR babies OR nursery) OR nurseries AND TITLE-ABS(attention OR joint attention) AND TITLE-ABS(child development OR infant development). This search produced 3,864 results; Embase ‘Infant’/exp OR ‘neonate’:ti,ab OR ‘newborn’:ti,ab OR ‘newborns’:ti,ab OR ‘baby’:ti,ab OR ‘babies’:ti,ab OR ‘nursery’:ti,ab OR ‘nurseries’:ti,ab AND ‘attention’/exp OR ‘joint attention’:ti,ab AND ‘child development’/exp OR ‘childhood development’:ti,ab OR ‘infant development’:ti,ab. This search produced 7,801 results; PsychInfo (DE “Infant Development” OR DE “Infant Temperament” OR DE “Neonatal Development” OR infan* OR neonat* OR newborn OR newborns OR baby OR babies) AND (DE “Selective Attention” OR DE “Sustained Attention” OR (DE “Attention” AND (DE “Social Behavior” OR DE “Joint Attention”)) OR (joint NEAR/7 (attention))). This search produced 991 results. Overall, a total of 13,898 articles were produced from the four databases.

2.1.3. Study selection

All 13,898 articles were imported into the online screening tool Covidence software (2022). The software removed 1,748 articles for being duplicates which left 12,150 articles for title and abstract screening. This process involved two separate screeners (HB & AM) to scan through all of the article title and abstracts and decide if they were relevant to the question by pressing either yes, no or maybe. In order to avoid bias, all voting was blinded, meaning that each screener was unaware of the other’s vote. If an article receives two no votes then it is classed as an irrelevant reference, if an article receives two yes or two maybe votes then it is moved to the next screening stage, and if an article receives one no vote and one yes or maybe vote then it is moved to a separate stage in which the screeners discuss and come to a joint decision on whether the article is relevant and thus goes to the next screening round. As a result of this process, a total of 486 articles were left for the full screen review. The full text review of 486 studies was completed by the two screeners and as a result of this, 106 articles were left for data extraction. These 380 studies were excluded for not fully meeting the eligibility criteria. For example, the papers did not use an explicit measure of JA, or the participants in the study were older than 12 months of age. See Figure 1 for the Prisma Flowchart for this scoping review.

Figure 1:

A PRISMA flowchart to illustrate the study selection process.

2.2. Data charting and extraction

The Covidence data extraction template was modified in order to meet the needs of the current scoping review (e.g., including what JA measure was used in the paper). A data extraction table was independently created by the lead author (HB) for each of the 106 articles. The data extraction form captured the following characteristics: (1) Title; (2) Authors; (3) Country in which the study was conducted; (4) Aims of the study; (5) Population description; (6) Total number of participants; (7) JA assessment tool; (8) Any relevant notes. In order to verify that the data extraction table was consistent across the papers, 25% of the final extractions were randomly spot-checked and approved by a second reviewer (LK).

3. Results

3.1. Study characteristics

A list of the 106 studies included in this scoping review are presented in the supplementary information. All studies were published between 1986 and 2021, with 23 studies published in the last 5 years. In addition, all of the studies were experimental designs. Studies were conducted internationally, with The United States being the country from which most research was published, but also included research from Europe, including Germany, The United Kingdom, Belgium and Portugal, and Asia, including Japan. Sample sizes for the experiments ranged from 12 to 325 infants, with the mean number of infants included across studies being 65 (SD =60.9). The age range for inclusion in this scoping review was infants under 1 year of age, participants were assessed for JA skills in these studies from 1 to 12 months of age.

Studies assessing both typically and atypically developing infants were included in this scoping review. Although within the papers the majority of the infants were TD (n=85), papers also included pre-term infants (n=7), infants with chronic otitis media (n=1), infants with pre/perinatal brain lesions (n=1), infants at elevated likelihood for autism spectrum disorder (n=9), infants with downs syndrome (n=2) and infants prenatally exposed to cocaine (n=1).

A range of different measures was used to measure JA across the 106 papers. The mean sample size of participants used across JA measures varied: Data coding (M=47, SD=30.7), the Early Social Communication Scale (ESCS) (M=82, SD=76.8), eye-tracking (M=45, SD=30.8), questionnaires (M=83, SD=74), EEG (M=56 SD=29.6), other standardized measures (M=88, SD=95).

Figure 2 provides the distribution of assessment measures identified within these papers. Within this figure, the distribution of methods is divided into early and recent studies i.e. before and after the year 2000 in order to present the effect of technology on research trends. The earlier studies (pre-2000) used only qualitatively defined measures of JA. The accessibility of eye-tracking and EEG most likely increased in the period post-2000 explaining the increase in quantitative measures used to measure JA.

Figure 2:

A bar graph to show the distribution of joint attention measures used across research papers.

3.2. Types of study and main outcomes

All of the studies included in the scoping review were human subjects research studies. Within the studies, JA was assessed on its own or as a part of a battery of tools aiming to evaluate infant social development. Based on the data extraction, the identified measures were divided into qualitative and quantitative measures and included the following: Data coding, Questionnaires, Eye Tracking and Electroencephalography. A total of 106 papers were identified, all JA measures that were used will be reported; however, only a subset of the papers will be discussed in detail. We grouped the measures into two categories, qualitative and quantitative assessments of JA. Qualitative assessment included data coding and questionnaires. Although these measures quantify JA using some standardized method, an assessor or psychologist are needed to determine the nature or level of JA and thus this includes some subjective interpretation. Alternatively, the quantitative measures included eye tracking and EEG. These methods are more objective. A total of 95 qualitative and 11 quantitative methods were identified in the 106 papers. Below, we describe the qualitative and quantitative methods by category.

3.3. Qualitative measures

3.3.1. Data Coding

The overwhelming majority of the measures used to assess JA development in the research papers discovered in this scoping review consisted of coding both in-person and video data. However, different approaches to this method exist. Typical measures included coding videos of TD infants at home playing with their mothers in a non-structured way (e.g., free play) (Bakeman & Adamson, 1986). Specific coded behaviors included, for example, when the baby was in a state of person engagement or object engagement. This video-based coding method has also been used with infants with autism spectrum disorder (ASD) (Osterling & Dawson, 1994; Palomo et al., 2022), in which family home movies were retrospectively reviewed and coded by an experimenter for social orienting behaviors, and also coding videos of free toy play sessions to assess JA skills in very-low-birth weight infants (Garner et al., 1991). Other methods to assess JA development in TD infants have included real-time in-person coding of infants naturally initiating and responding to JA in mother-infant dyad play sessions (Mateus et al., 2013; Nagell, 1997). This method of real-time in-person coding has also been used to assess JA development in infants with Downs syndrome (Landry & Chapieski, 1989, 1990) and to assess this in infants from a low socioeconomic status (SES) background (Mendive et al., 2013). Aside from just assessing how infants behave in naturalistic play sessions, researchers have also manually coded infant eye gaze in TD infants in order to measure their JA (Cleveland et al., 2007; Morales et al., 1998). This typically involves coding the duration of looking times to novel toys in different conditions: a JA condition in which the adult speaks about the toy while alternating their gaze between the toy and the infant, and an object only condition in which the gaze is alternated between the toy and a spot on the ceiling. McClure et al., (2018) investigated a novel way regarding how infant JA development can be assessed via the use of screen-mediated video chats; a semi-naturalistic observational method in which examiners recorded two videos of the interaction which were then synced and coded for behavioral interest.

The largest disadvantage with regards to data coding episodes of JA is the lack of standardization. Researchers define episodes of JA in line with their behavioral coding instructions, which vary across researchers, labs, and institutions. This is problematic when comparing findings across papers; behaviors classified into JA episodes in accordance with one coding manual may not be classified the same way if a different coding manual was employed. However, a flexible data coding analysis may be beneficial when measuring JA in some scenarios dependent upon the session being coded. For example, when coding a naturalistic free-play session an exploratory analysis can be employed to assess some JA behaviors that have not been explicitly classified in the coding guidelines. This is especially relevant when assessing JA in atypical populations in which how JA manifests and presents is unclear, such as in infants with visual impairment. Flexible data coding also allows for different environments to be employed, and the opportunity for researchers to be present or not; manipulating these variables may be necessary dependent upon what research question is being asked. For example, a free play session recorded in an infant’s home is more likely to provide us with the most naturalistic JA bids from the infant. A further advantage of this type of measure is that it can provide researchers with evidence of both RJA and IJA, allowing us to assess JA as a whole.

3.3.2. Standardized Measures

There are also standardized assessments of JA that are often video recorded to code post evaluation. The most commonly utilized assessment within the literature is the Early Social Communication Scale (ESCS; P. Mundy et al., 2003; P. C. Mundy et al., 1984). The ESCS was initially developed as a clinical measure in order to assess a child’s social communicative skills. It is a videotaped structured observation based upon two constructs: 1) a Piagetian developmental stage-related orientation and 2) a pragmatic-functional orientation of goal-driven behavior. The scale involves a set of situations devised with the intent of encouraging interaction between an adult experimenter and an infant via specific toys and materials. One fundamental part of the ESCS is to measure specific JA behaviors. The ESCS was used across population types in the findings of this scoping review. It was shown to be successful in measuring JA in TD infants (Thorp, 2007; Venezia et al., 2004), infants with pre/perinatal brain lesions (Šimleša et al., 2007), preterm infants (Mateus et al., 2020; Olafsen et al., 2006, 2012), infant siblings of children with ASD (Cassel et al., 2007; Gangi et al., 2014; Ibañez et al., 2013), infants from a low SES (Willoughby, 1999), and infants prenatally exposed to cocaine (Mallik, 2001).

Other less common, but still used, behavioral measures exist in order to assess the development of JA in the first year of life such as the auditory and multimodal JA skills assessment (Adamson et al., 2016; Adamson et al., 2021). This is a relatively new measure that considers both multimodal and auditory RJA and IJA. Auditory JA skills refer to the infant’s ability to attend to a label in their vocabulary without the deictic acts of pointing or gazing in order to give a more wholistic overview of JA. This has thus so far only been used in infants with TD. Furthermore, studies have used the Dimensional Joint Attention Assessment which is a coding scheme devised to score a naturalistic play-based assessment. This has been used successfully with TD infants (Reilly et al., 2021) and infants at high risk of ASD (Stallworthy et al., 2022). Three other papers used a coding scheme devised by Bakerman & Adamson (1984). This scheme involves a standardized coding model of different states of JA including person, object, onlooking and passive joint engagement. All of the 3 papers using this measure to assess JA in infants under 1 year of age investigated only infants with TD (Bigelow et al., 2004; Rollins & Greenwald, 2013; Simaes et al., 2022). Further, The Naturalistic Observation Schedule of Infant/Toddler Behaviors (NOSIB) was developed solely for coding JA in a more quantitative manner (Poon et al., 2012) and was used successfully with children with ASD. The NOSIB codes the quantity of JA and imitation behaviors in order to reflect the frequency of which infants are engaging in social-communicative behaviors. Elison et al., (2013) aimed to identify how frontolimbic neural circuity predicts JA in TD infants by using an adapted measure of JA assessment (Deák et al., 2000) that focuses on eliciting naturalistic toy-based play with the intention of characterizing dimensional ratings of RJA to reflect individual differences in RJA performance. Other research may have also used individual combinations or adaptations of measures that cannot all be defined within this scoping review. However, it is clear that no one standardized behavioral coding approach exists.

The overarching advantages and disadvantages to this method are in-line with those of non-standardized data coding assessments. However, using a standardized measure has strengths in that it allows for cross-comparisons to be made across research papers employing it. This is useful in JA research as using standardized assessments in both typical and atypically developing infants allows for a deeper understanding of how JA manifests in different infant populations. Assessing both RJA and IJA may be possible when employing one standardized measure, such as the ESCS which has distinct scoring subsections for both RJA and IJA. However, some measures choose to focus on the elicitation and scoring of only one of these JA types. A standardized data coding and scoring system should be chosen carefully depending upon the type of JA that wishes to be assessed in the research question by the investigator.

3.3.3. Questionnaires

Two papers used report-style questionnaires in order to investigate early JA development in infancy (Koyama, 2016; Veness et al., 2014). Koyoma (2016) used a self-modified version of the Checklist for Development of Early Social Cognition (DESC; Morinaga, 2011) in order to assess Baka infants at multiple age points including 1,3,4,5,6,7,8,9,10 and 11 months of age. They report that no culturally neutral scale of early social cognitive development is yet to be developed and standardized. This measure involved asking the mother 22 yes/no questions with regards to five categories of social cognitive development, one being JA. The three categories for developing JA include: (1) Infant comprehends that parent is pointing at an object and looks in that direction, (2) Infant looks at objects that parents are looking at, (3) Infant tries to listen to the sound his/her mother is interested in.

Veness et al., (2014) aimed to identify social communication skills that act as predictors of autism at 7 years of age. Children with ASD, developmental delay, language impairment and TD infants at the age of 7 years had been assessed at 8 and 12 months of age. Participants were required to complete detailed questionnaires about their infant at recruitment (between 10–12 months of age) and then every birthday until 7 years of age. These questionnaires included the Communication and Symbolic Behaviour Scales (CSBS) Development Profile Infant-Toddler Checklist (Wetherby et al., 2003) which aims to assess early social, communication and symbolic behaviours (including JA) and the MacArthur-Bates Communicative Development Inventory: Word & Gestures. The CSBS aims to score the development of JA by investigating aspects including emotion and use of eye gaze, communication and gestures. Veness et al. (2014) found that significant predictors of ASD were found at 8 months of age, specifically concerning JA.

Questionnaires are especially useful methods when investigating the development of JA; this research typically requires a large sample size and therefore may benefit more from questionnaires over a period of several years. This JA measure also has the unique benefit of being able to gauge a parent or caregiver’s perspective on their infant’s current JA capabilities. This is useful as infants typically behave most naturally with their caretaker. Current questionnaires tend to involve binary yes/no questions which are limited in that they offer no scope for further detail and clarification. This may be most important when attempting to investigate JA development in atypical populations. In addition, questionnaires answered by a caregiver are subjective and they may not represent an infant’s true ability of RJA or IJA. Questionnaires are a useful tool, but it must be remembered that it is difficult for this one measure to obtain a true picture of an infant’s JA behaviors and so may best be used in combination with other measures depending upon the research question.

3.4. Quantitative measures

3.4.1. Eye Tracking

Eye tracking is a form of sensor technology which detects and follows what an individual is looking at in real time. Eye tracking uses infrared or near-infrared light sources along with cameras that track either one or both eyes (Autoridad Nacional del Servicio Civil, 2021). In the majority of modern-day eye tracking technology, an arrangement of discreet light sources cast light upon the eye creating a corneal reflection. The eye tracker records the relationship between the center of the pupil and this reflection in order to relate the eye position to locations in the world (Brunyé et al., 2019). This renders it as a potentially useful measure in order to assess infant JA as it allows the researcher to perceive their subtle shifts in visual attention over trials and time (Aslin, 2007). It has been claimed that using eye tracking over behavioral observations of JA enforces gaze measurement accuracy but reduces ecological validity (Ambrose et al., 2021). However, ecological validity can be improved by using head mounted eye-tracking. Seven of the included research papers for this scoping review chose to use eye tracking as their assessment of JA development in the first year of life (Boyer et al., 2020; Brandone, 2015; Gustafsson et al., 2015; Ishikawa et al., 2019; Nyström et al., 2019; Yu et al., 2019; Yu & Smith, 2016). All of these papers investigate TD infants except for Nyström et al., (2019) who included infants at familial increased risk for autism.

The majority of the research papers used screen-based eye tracking. Boyer et al., (2020) used a Tobii 2150 corneal reflection eye-tracking system with a 21.3-inch flat LCD screen. The infants sat on their parent’s lap while watching a video. This system enabled the researchers to track both of the infant’s eyes in order to observe where each infant was looking on screen while watching the videos. Similarly, Brandone (2015) used two different Tobii eye trackers (a 17-inch and a 24-inch screen), Gustafsson et al., (2015) used a Tobii X120 eye tracker with a 24-inch screen and Ishikawa et al., (2019) used a Tobii T60 eye tracker with a 17-inch TFT monitor in order to achieve the same goal. Nyström et al., (2019) used a Tobii TX300 eye tracker in order to record the gaze of the infant, this eye tracker was not screen-mounted but instead placed on a table in-between the infant and the stimulus. The stimulus in this experiment was a live researcher, and not an on-screen video.

Both Yu et al., (2019) and Yu & Smith (2016) used head-mounted eye tracking in order to assess infant JA. They used head-mounted eye trackers from Positive Science, LLC on both the parent and infant. This technology uses an infrared camera that is pointed at the right-eye of the participant in order to record eye image, this is coupled with a scene camera in order to capture the first-person perspective of the participant. Each eye-tracking system allowed the researcher to observe both the egocentric-view and participant gaze direction. Three studies (Nyström et al., 2019; Yu et al., 2019; Yu & Smith, 2016) used a live stimulus (such as an experimenter or researcher) as opposed to an on-screen video. Furthermore, one of these studies (Nyström et al., 2019) included infants at familial risk of autism.

It is interesting to note the difference in uses of eye-tracking to measure JA in the first year of life. Screen-based eye tracking has the disadvantage that it only allows a researcher to observe where the infant is looking at on the screen; however, infants do not follow explicit instruction and thus we cannot ensure that the infant’s focus of attention is maintained on the screen, as opposed to something else. However, screen-based eye tracking allows a suitable assessment of RJA and so would be appropriate to use in an investigation into this. On the other hand, it would be difficult to observe IJA in screen-based tasks and so screen-based eye tracking fails to provide an overview of both types of JA. Using eye tracking on an infant when they are engaged in a real-life socially entrenched interaction may potentially yield results that are more generalizable to everyday life and provide us with quantitative data with regards to both RJA and IJA. This measure may work most appropriately in infants with typical development who can engage in shared looking and eye gaze modulation. Without this ability, eye-tracking may fail to account for some infants’ JA capabilities. For example, infants with visual impairment may present RJA and IJA behaviors in different ways due to an inability to engage in shared looking; eye-tracking may fail to pick up on their unique JA behaviors including gestures and postural movement. Once again, it is clear that this measure has unique advantages and disadvantages to measuring JA in the first year of life but may be a suitable tool depending on the research question being asked and the infant population being investigated.

3.4.2. Electroencephalography

Electroencephalography (EEG) is a relatively inexpensive neuroimaging technique used for measuring brain function. EEG measures electrical activity resulting from the synchronous firing of large populations of neurons (Biasiucci et al., 2019). A human brain contains millions of neurons that are responsible for controlling behavior by carrying information between the brain and the body. The recording of this electrical signal is achieved by placing electrodes at the scalp surface of the individual. EEG has advantages over other neuroimaging techniques in that it has excellent temporal resolution; the neural activity can be measured at the level of the millisecond. EEG, in comparison to other neuroimaging techniques such as fMRI, has been criticized in the past for its poor spatial resolution; however, recent advances in EEG analysis are aiming to improve this (Burle et al., 2015). With regards to infant research, EEG is practical. EEGs can be done in the infant’s home, are non-invasive and can be completed somewhat quickly. This measure of the brain allows researchers to measure cortical function underlying certain actions or behaviors. Thus, EEG may work as a complement to behavioral measures of assessing JA development in the first year of life. Four of the included research papers for this scoping review chose to use EEG as their assessment of JA development in the first year of life (Kopp & Lindenberger, 2011; Nichols et al., 2005; Rayson et al., 2019; Striano et al., 2006). All of the studies’ participants were TD infants, and all studies differ in their usage and findings of EEG in order to capture measure infant JA.

Striano et al., (2006) investigated the negative component (Nc) of the Event-Related-Potential (ERP). They found an enhanced peak amplitude of the Nc when the infants viewed objects following live interactive JA sessions in relation to the Nc peak amplitude following non-JA. This would illustrate that joint-attention interactions benefit infants by aiding them to focus their attentional resources on an environmental target. Similarly, Kopp & Lindenberger (2011) familiarized infants with objects in two different conditions (high and low JA) and then recorded EEG in response to old and new objects after the familiarization phase (immediate recognition) and following a one-week delay (delayed recognition). In an ERP analysis, it was found that JA modulates the amplitude of positive-slow-wave activity in immediate recognition, whereas in delayed recognition JA modulated the amplitude of the Pb component. In both experimental conditions, an enhanced Nc component was associated with the processing of unfamiliar objects. Nichols et al., (2005) wanted to investigate whether left or right frontal brain activation was associated with initiating JA. In this study, initiating JA was described as the number of times the infant alternated eye gaze between the tester and the active toy. It was found that initiating JA is associated with decreased left frontal hypoactivation and greater right activation, in contrast to previous literature (Mundy et al., 2000). Finally, Rayson et al., (2019) used EEG to record infant cortical responses to observation of an adult looking at the same object as the infant (congruent) or turning to look at another object (incongruent). Greater suppression of alpha band activity in central and parietal neural regions was observed in the congruent condition in comparison to the incongruent condition, this congruency effect was stronger as the infants increased in age. All papers successfully used EEG to evaluate an underlying cortical/neural network for JA.

There is no standardized EEG measure of JA assessment, as illustrated by the findings of these four research papers. This may be perceived as a disadvantage as there is no uniform way of analyzing EEG data in order to establish comparisons across infant population groups (e.g. typical vs atypically developing). However, this can be achieved in future research. EEG complements behavioral assessments by allowing us to gauge an objective understanding of underlying brain function and helping us to understand the relationship between JA behavior and brain function. In addition, EEG can be a very useful tool in assessing how brain function and behavior changes as a result of intervention. With regards to JA assessment, EEG must be used alongside a behavioral measure to ensure that the infant is engaging in JA episodes. A behavioral measure must be coded, and then the EEG data must be analyzed at the time points in which the infant was engaged in either RJA or IJA episodes. EEG is useful is assessing both RJA and IJA, and is a tool that will allow us to gain a greater understanding of the distinction between the two. Thus, research interested in understanding the underlying brain function of JA should employ EEG alongside a carefully chosen behavioral assessment dependent upon the research question and type of JA that is to be assessed.

4. Discussion

The purpose of this scoping review was to survey the existing literature for methods used for assessing JA in the first year of life and to synthesize these methods into quantitative or qualitative categories. By doing this, we aimed to better understand the current state of the field and determine the strengths and weaknesses of the most frequently used measures. From this, we can guide future research on what method, or combination of methods should be employed when assessing JA in the first year if life. We found that a majority of studies utilized qualitative methods to assess JA. Qualitative methods varied greatly between standardized assessments, non-standardized assessments, and behavioral coding. As aforementioned, although the qualitative assessments still quantify JA using a standardized method, an assessor or psychologist is often needed to determine the nature or level of JA, and thus this includes a degree of subjective interpretation. In addition, across all method types the majority of infants included in the research were TD. Atypically developing infants in the first year of life were heavily underrepresented.

A limited number of papers (n=11) used more-objective and quantitative methods to assess JA in the first year of life; these methods included EEG and eye-tracking. The eye-tracking assessments varied between using screen-based and real-life stimuli in order to measure infant looking. Using a screen-based measure to assess infant JA brings into question the social generalizability of the measure, as episodes of JA are typically conducted in person. Further, the studies employing EEG allowed researchers to observe the underlying neurobiology of JA behaviors. Quantitative measures are objective, meaning they allow for removed bias and concrete variables for analysis. Thus, EEG may work as a complement to behavioral measures of assessing JA in the first year of life. A standardized behavioral assessment alongside a brain-based measure, such as EEG, would allow us to understand more about the relationship between JA behavior and brain function. This can be achieved in both typically and atypically developing infants in order to understand some of the underlying mechanisms of atypical JA development. Consequently, this can potentially be used to help to understand when to implement early intervention and whether the intervention may lead to improved developmental outcomes.

Limitations exist with scoping reviews. Scoping reviews are broad, and although this is useful for achieving a comprehensive overview of the literature, it can be difficult to gather all the relevant information. For example, some papers that were excluded may have measured relevant aspects of social competence, such as eye gaze, but not explicitly related this to JA assessment.

5. Conclusion

The purpose of this scoping review was to provide a comprehensive overview of the most common methods used for measuring infant JA in the first year of life, understand their strengths and limitations, and guide future research on best practices of JA assessment. Addressing these areas are important because JA is a foundational skill that is crucial in supporting the development of social competence. Infants who experience impairments in JA go on to having poorer language skills, social communication impairments, and cognitive delays in comparison to their peers with typical development (Adamson et al., 2019). Delayed, or atypically developing, JA is an early indicator of a range of developmental delays. Thus, measuring JA in the first year of life allows us to understand more about typical and atypical trajectories of infant development and can ultimately be beneficial for identification and consequent intervention for infants with delayed or atypically developing JA.

Methods for measuring JA should be flexible yet carefully chosen depending on the research aim of the investigator. For example, research investigating the development of JA longitudinally requires a large sample size and therefore may benefit more from questionnaires over a period of several years. In addition, screen-based eye-tracking may be successfully used in experiments measuring current capabilities of RJA, but this measure would not be appropriate in assessments of IJA. If researchers aim to assess how JA affects other parts of cognition in the short-term then it may be necessary to control the environment experimentally and elicit JA behaviors by experimenter’s manipulations, as infant behavior and noise may be too difficult to control in natural interactions. Finally, research may benefit from utilizing a combination of both quantitative and qualitative methods. For example, in order to understand the brain function of JA then we would need to employ both qualitative and quantitative types of methodology e.g. EEG and the ESCS. These methods can complement one another to provide novel insights into infant JA developmental trajectories, both typically and atypically developing, by helping us to better understand the underlying neural mechanisms of JA. A combination of subjective and objective measures in this instance would also be useful in assessing intervention change. Ultimately, appropriate methods should be employed in future studies focusing on JA assessment in the first year of life dependent upon the research question, study design, and environment. Subtle nuances in any of these three variables will affect the utility of the chosen JA measure, and the strengths and limitations of the chosen measure will vary depending on the research aim of the investigator. In addition, more studies on infants with atypical or delayed JA would improve the field’s understanding of the development and best practice assessment of JA across more diverse populations.

In conclusion, JA is a key developmental skill that underlies several different developmental areas including social competence, self-regulation, and general intelligence. Impaired JA early in life can be an indicator of a later emerging developmental issue and so assessing JA in the first year of life is important so that impaired or delayed JA can be identified, and early interventions implemented. Researchers interested in investigating JA as an early emerging social cognitive process should carefully choose their assessment measure, or combination of measures, in line with their research aim. We propose that utilizing a combination of methods, such as EEG and a behavioral assessment, would allow us to fill gaps in this field by learning more about the underlying neural correlates of JA that can inform future assessment and intervention.

Data Availability Statement

Data sharing is not applicable to this article as no new data were created or analyzed in this study.