Individual Differences in Mental Attentional Capacity Across Development: an Eye-tracking Study

Размещено на http://www.allbest.ru/

Individual Differences in Mental Attentional Capacity Across Development: an Eye-tracking Study

Introduction

Common sense tells us, that what we look at corresponds to what we are mentally focused on. Although this is not always true, eye-tracking is still very useful for studying attention (Richardson et al., 2007) and related cognitive processes. While eye-tracking is not a direct measure of brain activity, it has been shown to reveal information about mental processes, that may not be easily accessible through other measures, such as decision-making and problem solving strategies (for a comprehensive review on eye-movements in decision making see Orquin et al., 2013, for spatial attention see Kiefer et al., 2017). Another definitive advantage of eye-tracking methods is that this technique is non-invasive and can be used both with adult and children. Finally, due to the fact that eye-tracking is a relatively well-established field and many of its indices, such as saccades and fixations, are already well studied, more informed inferences can be made regarding theoretical implications of the results. Specifically a large body of eye-tracking research with children exists that uses basic pro-and anti-saccade tasks. This research is focused primarily on understanding eye tracking indices related to atypical development such as autism spectrum disorders (for related meta-analysis see Chita-Tegmark, 2016), dyslexia, dyscalculia and attention deficit hyperactivity disorder (see Rommelse et al., 2008 for a review of eye-tracking in developmental psychiatry, see Quercia et al., 2013 for a review in dyslexia, see Mahone, 2011 for a review of eye movement correlates of ADHD). Although research with children, particularly school-aged children, is limited relations among higher order cognitive abilities, such as working memory, and eye-tracking indices have been proposed.

The term mental attentional capacity was introduced by Juan Pascual-Leone in the Theory of Constructive Operators within the framework of neo-Piagetian approach to cognitive development, where it is also known as the M-operator (Pascual-leone, 1970). It corresponds to the amount of schemes (i.e., operative and figurative), that can be maintained and processed in mental attention and thus could be interpreted as a central component of working memory (Arsalidou et al., 2010). There is a further distinction between structural mental attentional capacity and functional, where structural corresponds to maximum potential capacity and functional - the capacity used at any given moment (Chapman, 1981). Mental attention itself can be seen as conceptual equivalent of executive attention, however it is further substantiated with quantitative, graded increments in processing complexity across development (e.g., Pascual-Leone & Johnson, 2005). Mental attentional capacity was hypothesised to increase linearly with age, reaching the amount of 7 schemes by 15-16 years, corresponding to the “magic” number 7±2, suggested by at Miller in 1956 (Miller, 1956) the beginning of working memory research (Pascual-Leone & Johnson, 2011). Several developmental studies have yielded evidence to support the hypothetical trajectory of mental attentional capacity development (see Pennings & Hessels, 1996; Arsalidou & Im-Bolter, 2017 for review). Mental attentional capacity or working memory capacity has been shown to be closely related to other cognitive abilities and intelligence and working memory dysfunction can be used as a indicator and predictor or various cognitive and neurological disorders, including developmental disorders (Simmering & Perone, 2013 for review).

Despite the fact that a review of eye-tracking studies of working memory-related processes in children would be highly useful in informing future research, we have encountered no such review. Prior to formulating the hypotheses and design of this study a review of existing research was therefore conducted.

A systematic literature search was conducted (PubMed, Web of Science, and Google Scholar) using key terms eye-tracking, eye movements, children, development, working memory. Articles found through this search were manually screened to identify and select studies for the in-depth review. We have then identified the tasks and working memory models that were used in these studies as well as the type of eye-tracking tasks and eye movements' indices used. Source information of original study that were considered in the review, such as the age of participants, task, and recorded eye-tracking indices are tabulated (Table 1).

1.Models of working memory used in eye-tracking research in school aged children

One of the best known and most prominent working memory models is the Multiple-Component model, first proposed by Baddeley and Hitch in 1974 (Baddeley, 2003). In this model working memory is theorised to consist of several components: visuospatial sketchpad, phonological loop, episodic buffer and a central executive component. The first two components are parts of the “storage system” in two modalities: auditory and visual. The episodic buffer has been introduced in the later versions of the model to account for the relation between working memory content and conscious experience and serves as a “global workspace”, where items from different modalities can be processed together. Finally, central executive is a control system.. The role of the central executive could be interpreted much like that of executive or mental attention: central executive ensures that task-relevant information is kept in working memory and inhibits task-irrelevant information from taking up working memory capacity.

Another prominent model was proposed by Cowan and modified by Oberauer (Cowan, 2010). The Embedded Process Model acknowledges and includes long-term memory (LTM). Working memory is considered to be a part of LTM with two components: activated LTM, which is virtually unlimited, and focused attention, which can maintain only a limited number of items. In this model executive processes are theorised as those that allow new representations to form from LTM connections and new information entering working memory in order to ensure task completion. Although the model was originally developed for adult research, recent studies (Cowan et al., 2011) have attempted to reinterpret it within a developmental paradigm.

2.Working memory tasks in eye-tracking research in school aged children

A measure of working memory capacity in developmental research used predominantly within the field of psychological assessment are forward and backward digit span tasks of the Wechsler Intelligence Scale for Children-Fourth Edition (WISC-IV). In the forward digit-span task a child is asked to repeat a set of numbers to the experimenter in the same order as the numbers were heard and in the backwards task, in the reverse order. The forward task precedes the backwards and the number of numbers in the set increases with each successful repetition up to 9 numbers (Ramsay & Reynolds, 1995).

Another popular task in working memory research is working memory updating task, also known as a running memory task, where a participant hears or watches a sequence of items and is asked to remember the most recent item from the set (Bunting et al., 2006). The length of the list changes from trial to trial and the participants do not know in advance what length the list is going to be. At the end of the trial the participant is presented with a randomly chosen item from the set and is asked to compare that item with the one they remember and determine if they are the same or different. One disadvantage of this task is that despite the fact that the participant is supposed to try to remember every item on the list since its length is unpredictable, they still only need to recall the last item, which complicates interpretation of difficulty levels.

Measures of mental attentional capacity that ensure linear increase in task-difficulty across levels without increase in need for executive control have been developed specifically for use in developmental science within the framework of the Theory of Constructive Operators (Pascual-Leone, 1970). Parametric measures developed by Arsalidou et al. (2010), specifically the Colour Matching Task with 6 levels of difficulty and two versions (Balloons/Clowns) used for mental attentional capacity assessment satisfy several important requirements: parametric levels of difficulty (i.e., one additional colour for every difficulty change), unchanging executive goal across difficulty levels and culture fairness (e.g., visual-spatial stimuli). In both versions of the task the goal of the player is to compare the colours they see in the picture to the colours they saw in the previous picture and press a button to indicate whether the relevant colours are the same or different as quickly and as accurately as possible, disregarding the position of the colours. The levels of difficulty correspond to number of relevant colours to hold in mind and compare, plus a constant for executive schemes needed in each version of the task (Balloons: n + 1 and Clowns: n + 2, where n is the number of relevant colours). Two colours (green and blue) are irrelevant and serve as distractors, since participants are instructed to ignore them. In the first version of the task a number of balloons appear in the picture. The Clown version of the task is proposed to be more difficult with inclusion of interference due to the fact that participants need to additionally ignore the information about which item of clothing the colours were associated with (see examples of stimuli in fig.1), which requires executive inhibitory control. Therefore, CMT can be used to access both developmental increase and individual differences in mental attentional capacity and examine strategy use in tasks with high and low levels of interference.

3.Eye tracking tasks used with school-aged children

Some working memory tasks have been adapted specifically for eye-tracking research. The oculomotor delayed response task (ODR) was first introduced in animal studies in 1989 and is used to access visual-spatial working memory (Funahashi et al., 1989). The participant is asked to remember the location of a target stimulus, which appears on the screen for a short period of time, while the participant is fixating on the centre of screen. After a varying delay (1-8 seconds) the fixation stimulus disappears and the participant needs to move their gaze to the remembered location. Different eye-tracking measurements such as saccade accuracy, peak velocity and latency can be recorded and analysed to make inferences about working memory performance. It is worth noting that unlike many working memory tasks ODR task tests the individual's capacity to encode and maintain information without manipulation.

A simple eye-tracking measure of executive control and inhibition is the fixation task. The participant fixates a central cross. Once the cross disappears, the participant needs maintain central fixation while distractor stimuli appear on the periphery for two seconds. Two types of errors can be recorded in this tasks: inhibition errors, when the participant makes a saccade towards the distractor and corrected inhibition errors, when the look towards the distractor is corrected within 1000 ms. The latter are generally seen as an indicator that the task was fully understood by the participant, but inhibition was not possible. Another measure is the proportion of directional errors, when participant also loses fixation, but the saccade is made in the opposite direction from the distractor.

Finally, two widely used eye-tracking tasks that are generally accepted as a measure of inhibitory control are prosaccade and antisaccade tasks. In both antisaccade and prosaccade tasks participants are instructed to look at a fixation cross in the centre of the screen until a cue is presented. In antisaccade (AS) task the participants are instructed to make an eye movement as fast as possible in the opposite direction from the cue, while in the prosaccade task (PS) participants are instructed to look at the cue location. Thus in AS the participant needs to engage executive attention in order to not look at a salient stimulus and the reaction times and direction errors in the task can provide a lot of information about cognitive processes. (Munoz & Everling, 2004). Some studies in adults have found evidence that performance on anti-saccades may be dependent on working memory capacity (Unsworth et al., 2004, Meier et al., 2018). Normative development of anti-saccade performance is quite well studied (Klein et al., 2001, Fukushima et al., 2000). In a sample of 8-25 years old (Kramer, Gonzalez de Sather and Cassavaugh 2005) found that mean anti-saccade latencies improved between the ages of 8 and 16 years and proposed that this change is supported by the maturation of the visual system of the brain. Luna and colleagues (2008) summarized studies of oculomotor control development with the conclusion that while most aspects of oculomotor control system are mature by childhood (school-age), the speed of information processing and generation of voluntary saccades reach maturity by mid-adolescence and abilities to retain as task set and to monitor performance continue to develop beyond adolescence. Hutton (2008) reviewed literature connecting cognitive processes to performance on popular saccade tasks, including the pro and anti-saccade tasks. The review noted the weakness and lack of replication success of correlations between saccade tasks and working memory and attention performance and stressed the importance of identifying the sources of enormous individual differences in performance on these tasks in healthy participants as well as standardising the measurement of performance in these tasks, such as error rate in anti-saccade.

4.In-depth qualitative review of eye-tracking studies of working memory and related processes in school-aged children

A very small number of studies have investigated the relation between eye-movements and cognitive processes related to working memory and inhibition. Due to the complexity of the topic, the majority of the studies selected for the review were conducted within different theoretical paradigms, accessing different but conceptually close constructs, such as interference skills, inhibitory control, executive attention, and cognitive control. While the small number of studies prevents us from making reliable generalisations, it allowed us to conduct an in-depth review of each study.

A study by Luna et al., (2004) used oculomotor tasks to characterize the maturation of three separate constructs related to cognition: processing speed, inhibitory control, and working memory. They recruited 245 participants from 8 to 30 years of age. Processing speed was measured with the visually guided saccade task, which requires a participant to fixate to a suddenly appearing stimulus; a process guided by bottom-up perceptual attention. Inhibitory control was accessed with an anti-saccade task. Finally, working memory, specifically visuo-spatial working memory was operationalized as performance on an oculomotor delayed response task. Analysis of the results showed that, for the ODR task, age was significantly related to accuracy of both the final fixation (which corresponded to the remembered location) and the initial saccade, which indicates that working memory capacity improves with age. The authors also concluded, that while the three processes in question develop concurrently and along similar trajectories, their development is for the most part independent from one another.

Notably, Eenshuistra and colleagues (2007) examined the relation between development of working memory and inhibitory control by using antisaccade paradigm. The study considered two rivaling hypotheses. The first hypothesis stated that an increase in working memory capacity accounts for developmental improvements in AS task and the second hypothesis was that maturation of inhibitory control accounts for the change in performance. The authors stress the fact that working memory and inhibitory control are generally viewed as being connected, yet dissociable and the nature of their interdependence is unclear and still a point of discussion between various models, namely Multiple Components and Embedded process models. Participants were 28 children aged between 8 and 13 (22 girls), recruited from a local primary school, and 21 young adults (21 female), who were undergraduate students. Data from several participants of their original sample were discarded and data from a total of 50 participants were chosen for the analysis. Individual working memory capacity was determined through forward and backward digit span tasks of the WISC-IV. Participants were required to perform eight tasks - an antisaccade and prosaccade tasks in four different conditions: a gap and an overlap fixation-offset conditions, a working memory update condition, and a control condition. In the working memory update condition, a concurrent working memory task was added to the typical AS and PS tasks. In the working memory updating condition participants were asked to perform the PS and AS tasks as well as remember the most recently presented item from the list in each trial and at the end of the set was asked to respond whether a randomly selected image from that set was the same or different than the most recent one. The stimulus set was presented for the duration of 417ms after the cue. In control and working memory conditions, the cue was presented immediately after fixation, while in gap and overlap condition the cue was presented 217ms after the fixation and overlapped with the fixation cross for 217ms respectively. The authors concluded that smaller working memory capacity in younger children explains age-dependent differences in AS inhibition task better than inefficient inhibitory control mechanisms: increasing difficulty of the working memory task and corresponding addition to working memory load decreased performance on the AS task in younger children, but not in 12-13 year-olds and young adults (Eenshuistra et al. 2007).

The latest review of eye-tracking in children in normative and atypical development was conducted in 2007 and summarized all existing research of saccades, smooth pursuit eye movements and pupil dilation in developmental science (Karatekin, 2007). The review showed that the largest body of research was dedicated to graphing the development of the oculomotor system itself. Only a few studies attempted to relate eye-movements to cognitive processes in children at the time, most related to visual-spatial attentional processes. Since 2007 most eye-tracking research has been conducted in atypical development (see references for relevant reviews in the introduction). Critical, several studies have investigated relations between eye-tracking indices and working memory or executive function in normative samples.

Roebers, Schmid and Roderer conducted two eye-tracking studies of working memory-related processes in a sample of primary school children in 2010. The first study (Roebers et al., 2010a) explored different encoding strategies in children between 7-10 (7-8, N= 49, 9-10, N=51). The participants completed three tasks: encoding strategy task, distractibility task and flanker task, while their eye-movements were recorded. The encoding task consisted of two trials: baseline and critical trial. The critical trial differed from the baseline trial through the presence of a distracting stimulus in a list of images to be recalled. Results of the study revealed that recall was correlated with gaze time. Recall and encoding were more closely related in the 9-10 years group compared to 7-8 group.

The second study investigated the relation between working memory and interference control skills in a sample of 75 six-year-old children (Roebers et al., 2010b). In a between-subject design, children in the control group had to complete a Backwards color recall working memory task, while the experimental group had to complete the same working memory task with the presence of distracting stimuli. Fixation durations and location were recorded in both cases. In the Backwards color recall task, chidren were presented with a cover story, where a dwarf walking through a forest loses colored disks (different color circles), presented one after another on the screen (as digits in a classical backward recall test). The disks have to be “collected” - recalled in a reverse order. In the experimental condition distracting stimuli - images of objects, which could be found in a forest, appeared on the screen in addition to the colored disks. Children were shown the stimuli beforehand and specifically instructed not to attend to them. Importantly children also completed an Interference control task (an adapted Flanker task) before the main experimental task and were categorized into four groups based on their interference performance. The results showed a significant correlation between interference control skills and working memory performance as well as poorer working memory scores in the experimental condition. Interestingly, analysis of fixations in the interference condition revealed that the amount of time spent looking at distractors did not differ for children with high and low interference skills and did not affect performance. Authors propose that this finding could be explained by a more complex relation between eye-movements and visual attention.

A recent study investigated development of executive functions, such as inhibition and switching in preschool and school-age children. Mainville et al., (2015) compared data from neuropsychological tests of inhibitory control ability with data from eye-tracking tasks. A sample of 46 participants, from 5 to 8 years of age, completed three behavioral tasks, commonly used in clinical psychology settings. Specifically, the Walk-don't-walk task from Test of everyday attention for children (TEA-Ch), the well-known Tower of London task (ToL) and Knock and tap (KT) task. In the WDW task, a participant is presented with a sheet of paper with a row of squares. When the participant hears a regular tone, they need to draw a line in a square and when a different tone is heard, they need to refrain from drawing a line. The ToL task is usually used to access planning and strategy abilities, however, rule breaks score can also be used to access inhibition abilities (with a score of 0 being the best result). KT is a measure of motor inhibition, where a participant has to inhibit imitation reflex (the participant is required to tap the table with a closed fist when the experimenter taps it with an open hand vice versa) for the first half of the trials and then to inhibit both imitation reflex and a learned rule during the second half of the trials (when rules change and the participant needs to hit the table with a side of their hand when the experimenter uses a closed fist, with a closed fist when the experimenter uses a side of their hand and do nothing when the experimenter uses the palm of their hand). For eye-tracking measures of inhibitory control, the authors used fixation and anti-saccade tasks. Of the three neuropsychological tests, only WDW showed significant sensitivity to development. Percentage of correct antisaccades, and percentage of uncorrected errors correlated significantly with age on the antisaccade task. On the fixation task percentage of uncorrected errors and percentage of correct also showed a significant correlation with age. Statistical analysis of data from different tasks revealed that percentage of fixation directional errors and percentage of self-corrected fixations correlated significantly with WDW scores. After correction for age, the correlation between self-corrected fixations and WDW scores lost significance, indicating that the correlation could be explained by developmental changes.

Table 1. Review of eye-tracking studies of working memory in school-aged children

 Overall, research into relations between eye movements and mental attentional capacity across development is sparse and fragmented, due to differences in theoretical approaches and use of unique combinations of eye-tracking indices and working memory tasks. Only a few of the studies analyzed eye movements directly during a task designed to measure working memory specifically. The majority of the studies analyzed working memory performance separately and correlated it with eye tracking measures of other executive functions, most popularly inhibition. Importantly, research goals of the studies also differed. Some studies attempted to trace the development of working memory throughout childhood with the use of eye tracking, while others researched how much interference control and working memory capacity contribute to working memory eventual performance. Others still were interested in how well eye tracking and other types of measures capture working memory development. Notably, no studies have focused on looking for potential eye tracking indices of encoding and maintenance strategies or the change is eye movements with incremental increase in working memory load. To conclude, much further research is needed to broaden our understanding of the complex relationship between working memory in development and eye movements. Moreover, to our knowledge, no eye tracking studies have been conducted so far with parametric developmental measures, such as the Colour matching task, which would allow to dissociate changes in saccades and fixations related to cognitive load from those related to interference control and trace the maturation of these two processes.

5.Experiment

Hypothesis

We expected that increases in task difficulty, corresponding to an increase in mental attentional demand, will be associated with an increase in number of fixations (for use of saccades and fixations as a measure of mental effort see review Eckstein et al., 2017) and a corresponding decrease in duration of fixations for adults. For children, we expected the number fixations to increase as a function of difficulty similarly to adults. Additionally we expected to see age differences between adults and children, specifically we expected lower M-scores, slower reaction times and increased number of fixations for children as compared to adults. Effect of interference was also expected for both groups, with a larger number of fixations and shorter fixation duration in the high interference (clowns) condition.

Methods and participants

Participants

Adult study

Data are reported on 33 participants (mean age = 23±4.1, range 18-35, sd = 4.1, 17 male), with normal or corrected vision with contact lenses and no prior self-reported head trauma or cognitive impairments were recruited through advertisement for research participants on university campus and through the HSE mailing list. The original sample was 35 participants, however two did not complete both pro- and anti-saccade tasks and were excluded from analyses.

Children study

Fifteen participants were recruited for the children study. Data from 3 participants had to be excluded from preliminary analysis due to technical problems (e.g., ) during data collection. Current analysis included data from 12 participants (from grades 3-4, 9-10 years, 3 male), with normal or corrected vision and no prior self-reported head trauma or cognitive impairments, recruited from a pool of elementary school students in Moscow.

Adult participants and parents of child participants provided signed consent form. All methods and procedures were approved by the local ethics committee at the National Research University Higher School of Economics.

Eye-tracking measures

Colour Matching Task (CMT):

Parametric measures developed by Arsalidou et al. (2010), the Colour Matching Task with 6 levels of difficulty and two versions (Balloons/Clowns, example of stimuli in Figure 5) were used for mental attentional capacity assessment. A total of 24 task blocks is presented. Task block contains 4 runs of 6 blocks (32s duration), which contains 8 stimuli and only one difficulty level with blocks of different levels placed in pseudo-random order. Every task block ends with a baseline block (when no response is required, since only irrelevant colours will be present). Presentation of the stimuli will last for 3 seconds, during which the participant needs to make a response (same/different). Every participant completed all 24 blocks and reaction times and accuracy were recorded. Accuracy was calculated as percentage of correct responses for the trials that required a response for every level of difficulty and levels with 70% of correct responses or more were considered passed and included in the individual mental attention (M-) score (i.e., M-score). Interference effects were considered in calculation of individual M-scores of participants: 1 was added to the final score for the balloons and 2 for the clowns condition.

Figure 5. Example of CMT stimuli (balloons and clowns condition)

Pro and Anti-saccade:

Classical antisaccade and prosaccade tasks were added to the experimental paradigm as a potential way to control for individual inhibition differences in participants as well as a control task to ensure that the sample is normal in terms of eye tracking data. Based on existing literature, the latency of the first correct saccade, which is calculated as the elapsed time between the onset of the target and the generation of a saccade in the correct direction, was chosen as the defining variable for assessment of performance in the pro/anti-saccade task.

Experimental procedure

Participants were invited to complete a series of trials in a behavioural task after a detailed scripted, verbal explanation of the task. During this task their eye-movements were recorded with the use of The EyeLink Portable Duo (SR Research) recording at frequency 1000 Hz in remote head-free-to-move mode. Prior to each experimental task for both versions of CMT, calibration for the pupil was done for 9 points and colour-blindness was ruled out for all participants. After completing the task participants were also asked to complete a short questionnaire, detailing their strategies for performing the task successfully. The behavioural task was either directly followed or preceded by antisaccade/prosaccade tasks (to ensure no effect of tasks on each other).

The adult participants completed the task in the laboratory while the children completed the task in a quiet room at their school.

Data on fixations was extracted using EyeLink Data Viewer software and analysed using R packages for statistical analysis.

Results

Adult study

Accuracy and reaction times

Proportion of correct responses (accuracy) and response times for all levels of difficulty, as well as M-scores were calculated for both versions of CMT (balloons and clowns; Tables 2a and 2b; Figures 1 and 2). Only trials with correct responses were selected for the analysis of response times. Mean M-scores for CMTs were calculated: mean for balloons is 5.6± 1.06, mean for clowns was 5.4 ±1.4.

Effects of difficulty for accuracy and reaction times.

Accuracy decreased with increase in mental attentional load as expected and this decrease was steeper for the interference condition (clowns). ANOVA was conducted to test the effect level of difficulty on accuracy and revealed significant differences (F = 85.44, p < 0.001 , зІ = 0,48368969). An inverse trend was observed for reaction times on the successful trials (i. e. trials when participants provided the correct response) and ANOVA revealed significant differences for level of mental attentional load on reaction times (F = 1755.54, p < 0.001, зІ = 0,16489).

Effect of interference for accuracy and reaction times.

ANOVA was conducted to test the effect of interference condition on performance and revealed significant differences between low and high interference conditions (F = 32.49, p < 0.001, зІ = 0,03679124). ANOVA was additionally conducted to test the effects of interference on reaction times of participants and revealed significant differences (F = 187.93, p < 0.001, зІ = 0,01739).

Table 2a Means and standard deviation of accuracy (proportion of correct responses) and reaction times for the adult group for CMT-Balloons

Table 2b Means and standard deviation of accuracy (proportion of correct responses) and reaction times for the adult group for CMT-Clowns

Fixation duration and number of fixations

Means and standard deviation were calculated for duration of fixations and number of fixations per trial for every level for both conditions of CMT (Table 4a and 4b) and are represented in Figures 3 and 4 (for both age groups). A pronounced upwards trend can be observed for the number of fixations, meaning that more fixations were made with increase in difficulty, which is consistent with the predictions of this study. A complimentary downwards trend is apparent for the average duration of fixation.

Table 4a Means and standard deviation of number of fixations and duration of fixations for the adult group for CMT-condition balloons

Table 4b Means and standard deviation of number of fixations and duration of fixations for the adult group for CMT-condition clowns

Effects of difficulty for fixations

A two-way repeated measures ANOVA was performed to test the effects of difficulty level and CMT version on amount of fixations participants made and fixation duration across difficulty levels.

Contrary to our predictions, no significant main effect of condition (i.e., Balloons vs Clowns) was observed in terms of fixations. However a significant effect of difficulty level for both number of fixations participants made (F = 50.05, p = 0.000286, зІ = 0.96744564) and fixation duration (F =28.081, p = 0.00115, зІ = 0.93941201) was found and was followed up by post-hoc Tukey HSD tests. For both variables significant differences were observed found between lower (level 1 and level 2) and higher levels (levels 4-6) of difficulty.

Figure 3. Average with standard error bars duration of fixation for adults and children in the CMT for 6 levels of difficulty for balloons (low interference, facilitating) and clowns (high interference) conditions

Figure 4. Number of fixations with standard error bars per difficulty for adults and children in the CMT for 6 levels of difficulty for balloons (low interference, facilitating) and clowns (high interference) conditions

Antisaccade/prosaccade results in the adult sample

Latencies in anti-saccade (M = 300 ± 118 msec) and pro-saccade (M = 216 ± 98 msec) tasks were as expected. ANOVA was conducted for the effect of task (pro and anti) and revealed significant differences for the adults sample (F = 67.83, p < 0.001, зІ = 0.51) Difference between pro- and anti-saccade latencies was also calculated as a measurement of inhibition failure for each participant. Mean difference is 84±41 msec, is normally distributed and falls within the generally observed range for pro/anti saccade performance (Hutton, & Ettinger, 2006).

Results of the children study

Accuracy and reaction times

Means and standard deviations for accuracy and reaction times were calculated for every level of difficulty for both conditions and are presented in Tables 3a and 3b. As with the adult sample, only response times on successful trials were included in the analysis. Mean M-scores for CMTs were calculated for the children sample: mean for balloons is 5.4± 0.96, mean for clowns was 5.3 ±1.1.

Effect of difficulty for accuracy and reaction times

Similarly to the adult sample, children performed better, committed fewer mistakes and responded faster to trials with lower levels of mental attentional load. ANOVA was conducted to test the effect level of difficulty on accuracy and revealed significant differences (F = 27.06 , p < 0.001 , зІ = 0.4635).

ANOVA also revealed significant differences for level of mental attentional load on reaction times (F = 28.069, p < 0.001, зІ = 1.038083e-02).

Effect of interference for accuracy and reaction times.

ANOVA was conducted to test the effect of interference condition for the children sample on accuracy and revealed significant differences between the two interference conditions (F = 9.360, p = 0.00269, зІ = 0.03205452 ). ANOVA was additionally conducted to test the effects of interference on reaction times of children and also revealed significant differences (F = 13.612, p = 0.000229, зІ = 5.066148e-03).

Table 3a Means and standard deviation of accuracy (proportion of correct responses) and reaction times for the children group for CMT-Balloons

Table 3b Means and standard deviation of accuracy (proportion of correct responses) and reaction times for the children group for CMT-Clowns

Fixation durations and number of fixations

Means, standard deviation and standard error were calculated for number of fixations and mean fixation duration per trial for every level for both conditions of CMT and are illustrated in Figures 3 and 4 and presented in Tables 5a and 5b.

Table 5a Means and standard deviation of number of fixations and duration of fixations for the children group for CMT-Balloons

Table 5b Means and standard deviation of number of fixations and duration of fixations for the children group for CMT-Clowns

Effect of difficulty on fixations

In contrast with data extracted for adults, the upwards trend for the number of fixations is increasingly gradual for difficulty levels 1-4 and plateaus after level 4. A complimentary downwards trend is apparent for the average duration of fixation. Interestingly the difference between clowns and balloons condition virtually disappears after level 3, while there remains a distinct difference in accuracies (in Figure 1) for levels 4 and 5.

A two-way repeated measures ANOVA was performed to test the effects of difficulty level and CMT version on amount of fixations participants made and fixation duration across difficulty levels for the children sample the same as for adults.

Similar to adult data, no significant main effect of condition (i.e., Balloons vs Clowns) was observed on the number and duration of fixations. A significant effect of difficulty level for both amount of fixations participants made (F = 5, p = 6.39e-05, зІ = 0.1409805) and fixation duration (F = 4, p = 0.000928, зІ = 0.1326054) was found and was followed up by post-hoc Tukey HSD tests. For the amount of fixations made and for duration of fixation significant difference was found between level 1 and all levels of difficulty (2 to 6). Unlike the adult data, there was a difference between levels 1 and 2, and no other significant differences between levels were found.

Antisaccade/prosaccade results in the children sample

Latencies in anti-saccade (M = 400.93 ± 174 msec) and pro-saccade (M = 237 ± 128 msec) tasks were calculated for the children sample. ANOVA was conducted to test for the effect of task condition (pro and anti) and revealed significant difference (F = 40.33, p < 0.001, зІ = 0.64). Difference between pro- and anti-saccade latencies was also calculated as a measurement of inhibition failure for each participant. Mean difference is 163.22 ±72.61 msec, is normally distributed as for the adult sample.

Comparison between age groups

Accuracy and reaction times

ANOVA was conducted to test the effects of age group on accuracy and speed of performance. While no significant difference between the two groups was found in terms of accuracy, significant difference was revealed for reaction times (F = 18.510, p < 0.001, зІ = 1.588661e-03).

Fixation duration and number of fixations

One-way ANOVA was conducted to test effect of age group on duration of fixations and number of fixations and revealed significant differences in both cases (p = 0.00412, зІ = 0.12605395 and p = 0.000217, зІ = 0.20617999 respectively).

Antisaccade/prosaccade performance

One-way ANOVA was conducted to test for significant differences between children and adults on latencies in the saccade tasks and revealed a significant effect of age group (F = 20.44, p < 0.001, зІ = 0.3222).

Discussion

I examined the effect of task demand and interference levels on eye movements in children and adults. There are three main findings: (a) Eye movement indices are affected by task demand such that the number of fixations increase and the fixation duration decrease. This effect is present in both children and adults, albeit children's indices levelled off at a lower level of difficulty, consistent with the lower mental-attentional capacity scores in this group. (b) When the mental-attentional demand of the task is within the individual's mental-attentional capacity fixation durations decrease by about 100msec, and this difference is eliminated when task demand is greater than the individual's mental-attentional capacity (i.e., when the individual is performing at chance). This difference is comparable for children and adults. (c) Although some difference can be observed between fixation correlates of performance in balloons and clowns, the difference did not reach statistical significance, no effect of interference is found. Overall, indices from parametric measures of mental-attentional capacity suggest that a trade off between task demand and fixation correlates exists and that this relation is graded up to about the difficulty level that is within the individual's mental attentional capacity. This is the first study that used a parametric protocol with six levels of difficulty in a higher order cognitive task to examine eye movements in children and adults. Results are discussed in terms of theories of cognitive development and methodological choices for task selection in future work.

Effect of task demand

Tasks of mental-attentional capacity have six levels of difficulty. Following theoretical predications an effect of task difficulty, corresponding to mental attentional load and interference level was observed both for adults and for children. Additionally a complimentary graded increase in response times was also observed for both groups and affected by interference condition.

As expected, effect of difficulty level is observed for both number of fixations and fixation durations. Specifically, number of fixations gradually increase with mental attentional demand and duration of fixations decrease complimentary for the adults group for both low and high interference conditions. The trends of fixation variables with regards to level of difficulty followed predictions and difference between levels is statistically significant between the highest and the lowest levels of CMT. This finding is consistent with the results of Chen et al. (2011), who found an increase in number fixations and decrease in duration of fixations associated with an increase in task demand in a visual task. As eye-movements are in part controlled by visual association areas in the brain, such as the precuneus and fusiform gyri, our result is consistent with previous research showing a bimodal engagement between lower and higher levels, at least in adults (Arsalidou et al., 2013). Thus, I proposed that lower levels of difficulty require less effort and less sustained attention and allow the participant to start “mind-wandering” while keeping their eyes roughly on the same fixation point.

Group differences

While no differences between groups were found in accuracy of performance, significant differences were revealed in reaction times between children and adults with children responding slower in both interference conditions. This indicates that the accuracy-speed trade off was more costly for children as they found the task more cognitively challenging, despite similarity in M-scores.

This conclusion is supported by the fact that significant difference is found in the number of fixations and the duration of fixations between the adult group and the children group. Children made more fixations, especially during the lower levels of difficulty. Additionally, for the children sample, the trend of increase in number of fixations is different: number of fixations increased sharply after the first level of difficulty and then remained more stable,

Interesting discrepancy in the trends observed for the number and duration of fixations in children and adults could be explained by children utilising less efficient visual-spatial strategies during task performance in levels with mental load (i.e., difficulty) closer to the limit of their mental attentional capacity and the fact that children found difficulty level 2 more challenging as compared with difficulty level 1, than adults. According to task analyses, level 1 and level 2 in CMT-Ballons corresponds to a mental attentional capacity of 3 and 4, whereas level 1 and level 2 in CMT-Clowns corresponds to a mental attentional capacity of 3 and 4 (Arsalidou et al., 2010). Therefore, level 2, particularly in the CMT-Clown as expected to be more difficult. Combined with the data on reaction times, this indicates that adults found each higher level of mental attentional load more effortful than the previous one and the highest levels (with more than 5 items) similar in difficulty; these were the items above the mental attentional capacity of our sample (i.e., M-score ? 5.5). Children, on the other hand, found all difficulty levels above their mental attentional capacity (3-4 items) similar in difficulty and performed at chance, whereas for lower levels they were able exerted their maximum effort starting form levels 2-3.

Significant different were also found between the adults and children in pro and anti-saccade performance with children taking more time to reorient to the target stimuli in the presence of a salient distractor. This finding is consistent with the predictions of the study and is supported by prior research into development of oculomotor inhibitory processes (Luna et al., 2008, Hutton et al., 2006).

Effect of interference on eye-movements

Interference in the mental-attentional capacity tasks is manipulated by having two different contexts in which participants have to extract colors either from a simple figure of a set of balloons or a more complex figure of a clown. Because of the difference in visual features of each type of figure it was expected that eye-tracking indices would be different between the high and low interference tasks as the higher interference (clowns) is more effortful requiring a larger number of fixations. Although some difference can be observed between fixation correlates of performance in balloons and clowns, the difference did not reach statistical significance and no effect of level of interference is found. A tentative explanation could be offered from accuracy and M-scores data that in the present sample of adults effect of increased mental attentional load (levels 4-6) interfered with condition effect, meaning that participants found higher levels of low-interference version of CMT roughly as difficult as higher levels of high-interference version.