Developmental trajectories of hot and cool behavioral control across
childhood: a longitudinal study
Name: Niyana N. Kwee
Student number: 10988327
Bachelor: Psychobiology
Daily supervisor: Mara van der Meulen, PhD Supervisor: Bianca van den Bulk, PhD
Date: 24-01-2020
Table of contents Abstract 3 1 Introduction 4 2 Methods 7 2.1 Participants 7 2.2 Prodecure 7 2.3 Experimental tasks 8 2.3.1 Stop-signal 8
2.3.2 Social Network Aggression Task (SNAT) 9
2.3.3 SNAT stories 10
2.4 Statistical analysis 11
2.4.1 Stop-signal 11
2.4.2 SNAT and SNAT stories 11
3 Results 11
3.1 Stop-signal 11
3.2 SNAT and SNAT stories 13
4 Discussion 16
4.1 Cool behavioral control 16
4.2 Hot behavioral control 16
4.3 Further research 17
4.4 Conclusion 18
Abstract
Behavioral control is one of the basic executive functions. It allows us to resist first temptations and adjust our automatic behavior to new situations. A distinction can be made between behavioral control in an emotional setting, i.e., hot behavioral control, and behavioral control in a non-emotional setting, i.e., cool behavioral control. Previous research found that the development of behavioral control starts at infancy and continues until early adulthood. Most research however did not make a distinction between the different forms of behavioral control. The aim of the current study was to investigate the developmental trajectories of both hot and cool behavioral control during middle childhood. Cool behavioral control was measured with the Stop-signal task. Hot behavioral control in a social setting was measured with the Social Network Aggression Task (SNAT), and a variation called the SNAT stories. The current study tested 256 participants (aged 7-9 during the first time point) on a yearly basis, for four consecutive years. Each year the Stop-signal task was conducted, the SNAT and SNAT stories were alternately conducted. A significant increase of cool behavioral control between the ages 7 to 10 was found. Results also suggest that the development of cool behavioral control levels off after the age of 10. No significant increase of hot behavioral control was found, although this result might be due to the use of two incomparable tasks. The results however do correspond with theories on the development of hot behavioral control. This study contributes to the knowledge on the development of different aspects of behavioral control. Further research should include more time points to examine the development of hot and cool behavioral control from childhood to adulthood.
1 | Introduction
Behavioral control, the ability to resist first temptations and not act impulsively, is one of the basic executive functions. It allows us to make well thought decisions and change automatic behavior so that we can adjust to new situations (Diamond, 2013). Executive functions, including (inhibitory) behavioral control, have been extensively studied in the past decades. Developmental studies showed that executive functioning emerges in infancy and rapidly develops during early childhood (Anderson, 2002). It continues to develop throughout adolescence and carries on until adulthood (Silvers et al., 2012; Zelazo & Carlson, 2012). Although the development of executive functioning continues until early adulthood, not every component of executive functioning follows the same developmental path. Anderson (2002) found different developmental trajectories for different aspects of behavioral control. While some components of executive functioning do not mature until early adulthood, behavioral-control and self-regulation seem to have matured by middle childhood (Anderson, 2002). However, most prior studies did not distinguish between different components within behavioral control and self-regulation. A distinction can be made between behavioral control in a non-emotional setting (i.e., cool behavioral control), and behavioral control in an emotional and/or motivational setting (i.e., hot behavioral control). Most studies have focused on the cool aspects of behavioral control, while few studies have taken into consideration the hot aspects of behavioral control. Moreover, little emphasize on the distinction between the developmental trajectories of cool and hot behavioral control has been made.
Cool behavioral control, that is, behavioral control in a non-emotional setting (e.g., withholding yourself from pushing the wrong button when playing a computer game) has been widely studied. Previous research mainly used the Go/NoGo task, Simon task and Stop-signal task (Diamond, 2013). Lewis et al. (2017) found, using the Go/NoGo task, a large developmental increase of response inhibition (i.e., cool behavioral control) between the age of 6 and 8 years, and a smaller developmental increase between the ages 8 and 10. Strommen (1973) used the children’s game “Simon Says”, in which children must inhibit their behavior when a specific signal is not presented, in different grades with children from the ages 4 to 10. She found that the amount of errors made decreased with each grade. The results from these previous studies suggest that cool behavioral control is developing between the ages 4 and 10 years, however, it also suggests that there is a difference in the slope of developmental growth between ages.
Just like the Go/NoGo and the Simon task, the Stop-signal task is extensively used in studies that focus on inhibitory behavioral control. It measures the time it takes to inhibit a planned reaction (the Stop-signal reaction time [SSRT]). This is done by alternating go trials and stop-trials (i.e., in which participants must inhibit their response). By varying the time between the go trial and the stop-signal, the minimum SSRT can be measured (Tillman et al., 2007). The Stop-signal is a highly suitable task for measuring the inhibitory behavioral control, especially in longitudinal studies, since it is adjustable to younger ages (Carver, Livesey & Charles; 2001), and thus can be used with different age groups. Therefore, multiple studies used this task to examine the development of cool behavioral control. Carver, Livesey and Charles (2001) used the Stop-signal task on preprimary (under 5.5 years), young primary (5 to 7 years), mid primary school children (7 to 9 years) and adults. They found that the reaction time decreased with age, as well as the amount of errors made, and thereby concluded that response inhibition improves with age. Ridderinkhof, Band and Logan (1999) similarly found a decrease in SSRT with age, using the Stop-signal task on a group of 6 to 8-year-olds, a group of 10 to 12-year-olds and a group of students (21.7 years on average). This is again evidence for the development of cool behavioral control with age. All together, evidence from multiple studies, using different tasks, show that the development of cool behavioral control starts at a young age and continues to develop until at least the adolescent years.
Somewhat less widely studied is “hot” emotional control. Whereas cool behavioral control is behavioral control in a non-emotional setting, hot behavioral control is behavioral control in an emotional setting (e.g., withholding yourself from lashing out at someone, when that person says something mean to you). Studies that investigated hot behavioral control mainly used the Iowa gambling tasks, the delay discounting task and, for younger participants, the Marshmallow task. In all three tasks participants can make the decision between a direct, small reward or a larger reward later on. Kerr & Zelazo (2004) used the Iowa gambling task to measure hot control in 3 and 4-year-olds. They found that 4-year-olds made more advantageous choices than 3-year-olds, suggesting that a developmental increase of hot behavioral control has occurred between the ages 3 and 4.
The tasks mentioned above measure the delay of gratification, a well-studied phenomenon of behavioral control. Delay of gratification is the ability to resist an immediate reward to obtain a larger reward in the future. It involves motivational and emotional reasoning, and is thus a suitable phenomenon to indicate hot behavioral control. Using a delay discounting task, De Water, Cillessen & Scheres (2014) found evidence indicating an increase of delay of gratification with age. They tested adolescents and young adults of the ages 12 to 27 years. Earlier on, Scheres et al. (2006) found a similar effect of age in an experiment with children and adolescents of the ages 6 to 17 years. Evidence from both studies suggest a development of hot behavioral control from early childhood until adolescence.
Casey et al. (2008) did examine the development of hot behavioral control between childhood and adulthood. They examined the brain development to explain the often risky, suboptimal behavior of adolescents, and observed an imbalance between the development of responsiveness to impulses, especially of socioemotional context, and the development of the control over this responsiveness. Their imbalance theory states that this imbalance arises when the development of the limbic system, responsible for the heightened responsiveness to socioemotional stimuli, increases faster than the development of the prefrontal cortex, important for the control over this heightened responsiveness. The difference in development is noticeable between childhood and adulthood, during the adolescent period, and can cause risky, suboptimal behavioral.
Although the tasks mentioned above are certainly more emotional than cool cognitive control tasks, there is one crucial factor missing: these tasks do not measure hot control in a social setting, which is an important factor in modern day. Part of the development of social skills nowadays is learning to react to social feedback. Receiving negative social feedback, lowers the self-esteem (Leary et al., 2003a), can induce sadness or anger and can even lead to aggressive behavior like cursing or physical aggression (Leary et al., 2003b; Twenge et al., 2001). Such negative reactions when reacting to social feedback can lead to poor social relationships. Behavioral control is thus a crucial part of social skills, and the other way around, a social component is crucial when examining behavioral control. Achterberg and colleagues therefore designed a task in which hot behavioral control is measured in a social setting, the Social Network Aggression Task (SNAT; Achterberg et al., 2016). Participants receive either negative, neutral or positive feedback from fictional same aged peers on a profile with personal questions. Participants were instructed to give a noise blast to each peer after receiving their feedback. The duration of the noise blast was used as an index of hot behavioral control. Relatively short noise blast after social rejection can be seen as high hot behavioral control.
It is clear that there are different forms of behavioral control. Still, little is known about the developmental trajectories of hot and cool behavioral control, and whether these might be different. Moreover, most research on behavioral control focused on early childhood (Anderson, 2002) and adolescence (Silvers et al., 2012), whereas middle childhood marks important changes in brain development as well as social interactions, both which might influence the development of
behavioral control. Although in early instead of middle childhood, Zelazo and Carlson (2012), as one of the few, did compare hot with cool behavioral control. They found that 3-year-olds had more difficulty choosing a delayed larger reward over a smaller immediate reward for themselves (i.e. hot behavioral control) than 4-year-olds did. Performance improved when they had to choose for the experimenter (i.e. cool behavioral control). This evidence suggests that both cool and hot behavioral control start developing before middle childhood. It also suggests that, at the age of 3 years, cool behavioral control is more developed than hot behavioral control. However, very few longitudinal research has been done on the development of behavioral control, and specifically on the development of hot and cool behavioral control as two separate components of behavioral control. This however, is of key importance in understanding the developmental path of behavioral control. Longitudinal research can give insights on individual changes over time, which are crucial in the understanding of developmental trajectories. The current study therefore focusses on the development of hot and cool behavioral control throughout middle childhood using a sample of 256 participants that was tested on a yearly basis, starting from the age of 7-9 until the age of 10-12 years. Cool behavioral control was examined using the Stop-signal task. The stop signal reaction time (SSRT) was used as an indication of cool behavioral control. A shorter SSRT time was identified as high cool behavioral control. Hot behavioral control was examined using the SNAT and a variation on the original, the SNAT stories. The duration of the noise blast in the negative feedback condition was used as an indication of hot behavioral control. A shorter noise blast was identified as high levels of hot behavioral control.
Previous studies suggest an increase of behavioral control with age, for hot behavioral control as well as cool behavioral control. Hot and cool behavioral control were compared within a single experiment and evidence was found that suggested that the development of both hot and cool behavioral control starts before middle childhood (Zelazo & Carlson, 2012). However, it is suggested that the development of cool behavioral control has its start before the development of hot behavioral control. The imbalance theory states that hot behavioral control slowly develops until adulthood, with an increasing slope for development near the end of adolescence (Casey et al., 2008). Based on evidence from previous research, the current study expected cool behavioral control as well as hot behavioral control to increase with age. Besides, it expected the development of both to continue until early adulthood. A developmental pattern of declining exponential increase of cool behavioral control, with a large increase of behavioral control over time at the start of development and a lower increase over time near the end of the development is expected for cool behavioral control. Moreover, based on evidence from Lewis et al. (2017) on the development of cool behavioral control that suggests a decrease in developmental speed specifically between the ages of 8 and 10 years, a decrease in developmental speed for cool behavioral control from the age of 8 years on was expected. Hot behavioral control is expected to increase with age as well, however, in contrary to cool behavioral control, a developmental pattern of exponential increase is expected. Moreover, during middle childhood, hot behavioral control is expected to develop at a slower rate. If the adolescent period would be examined, hot behavioral control would be expected to develop at a higher rate. Figure 1 schematically displays the expected developmental trajectories of cool behavioral control and hot behavioral control.
2 | Method 2.1 | Participants
Participants were selected from the middle childhood cohort (MCC) sample of the longitudinal twin study of the Leiden Consortium on Individual Development (L-CID; Euser et al., 2016). Families with same-sex twins born between 2006 and 2008, and living within a 2 hour travel from Leiden, were contacted by L-CID and invited to take part in the longitudinal study. Address information was obtained from municipal registries. Participants that were included had normal to corrected-to-normal vision, no psychiatric or neurological conditions that could have effected their performance and were fluent in Dutch. The initial sample included 256 families with a twin of the same gender in the age range of 7 to 9 years. To control for interdependence, only one participant of each twin pair was randomly selected. This left a sample of 256 participants at the first visit (MAge = 7.96, SDAge = 0.68, MIQ = 103.80, SDIQ = 11.57, percentage male = 48, percentage right-handed = 89). Table 1 provides an overview of the sample of the first visit.
2.2 | Procedure
For the longitudinal twin study of L-CID, participants were tested on a yearly basis, with home visits and lab visits alternating, 6 years in a row. For this study, data of the first four years were used. There was on average 1.09 (SD = 0.07) year between the yearly visits (T1-T2: 1.01 ± 0.05; T2-T3: 1.05 ± 0.09; T3-T4: 1.27 ± 0.17). Participants started the first visit at 7-9 years old, and were between the ages 10-12 years old during the fourth visit. The first and third yearly visit were conducted at the Leiden University Medical Center (LUMC). Participants performed, amongst other tasks, the Stop-signal task and the SNAT. The second and fourth yearly visit were conducted at home. Participants performed, again amongst other tasks, both the Stop-signal task, and the SNAT stories. The SNAT and SNAT stories were executed alternately, to prevent a learning curve for the SNAT. Figure 2 provides an overview of the set-up of yearly visits with alternating tasks.
Figure 1. Schematic display of the expected developmental trajectories of hot and cool behavioral control.
The development by age is plotted. The dotted line displays the expected development of hot behavioral control. The solid line displays the expected development of cool behavioral control.
Table 1.
Overview of sample of T1.
Age
(mean (SD))
IQ
(mean (SD))Gender
(% male)Dominant hand
(% right-handed) Sample after control forinterdependence (N =
2.3 | Experimental tasks 2.3.1 | Stop-signal
The Stop-signal task (adapted from Williams et al., 1999) consisted of one practice block of 25 trials, and four experimental blocks of 50 trials. Each trial started with a fixation point (1500-2000ms), followed by screen with an arrow, pointing towards the right or the left side. The arrows were initially presented in a green color. For stop trials however, the green arrow turned red after a variable amount of time. Participants were instructed to press the button on the right side of the keyboard when a green arrow pointing to the right was presented, and to press the button on the left side of the keyboard when a green arrow pointing to the left was presented. When an arrow turned red, participants were instructed not to press a button. The time between the green arrow appearing and the arrow turning red (i.e. the Stop Signal Delay (SSD)) was adjusted during the task, depending on the successfulness of the participant. When a participant successfully inhibited its reaction in a stop trial, the SSD in the following stop trial was prolonged with 50 ms. When a participant unsuccessfully inhibited its response to a stop trial, the SSD decreased with 50 ms. The default start of the SSD was set to 100 ms. In 75% of the trials, a Go trial was presented. The other 25% of the trials, a stop trial with an arrow turning red was presented (Figure 3). Participants were instructed to press as soon as possible, while making as little mistakes as possible. In between each block, the task would present the percentage of mistakes made by the participant, and the average reaction time. Research assistant reminded the participant to react as soon as possible, to make as little mistakes as possible, or to do both, depending on the amount of mistakes and the reaction time. This was done to ensure that the task was executed in the right way, and the participant was clear on the instructions of the task. The Stop-Signal Reaction Time (SSRT) was calculated by subtracting the mean SSD from the mean Go RT (i.e. the reaction time on Go trials) from correctly answered Go trials. The SSRT was used as a measure of cool behavioral control.
Figure 3. Design of the Stop-signal task.
Representation of a stop trial with duration of the different screens, and a go trial with duration of the different screens.
Figure 2. Overview of the procedure of the experiment.
2.3.2 | Social Network Aggression Task (SNAT)
The Social Network Aggression Task (SNAT) was designed by Achterberg and colleagues (Achterberg et al., 2016). Participants were asked to fill in a personal profile with questions about, for instance, their favorite movie, favorite food, biggest wish and one of their dislikes. These personal profiles were filled in at home, and handed in prior to the lab visit. Prior to the start of the actual task, instructions on the SNAT were given, participants were informed that their personal profiles were shown to other same aged peers, and that these peers had given feedback on their profiles. Pictures of the peers were shown together with their feedback. This could either be positive, neutral or negative. In case of positive feedback, the profile picture would appear on a green thumbs up. In case of neutral feedback, on a grey circle, and in case of negative feedback, the picture would appear on a red thumbs down (figure 4a). The pictures of same aged peers were, unknown to the participants, created by morphing together pictures of peers from an existing database. Participants completed a practice block of 6 trials, with each feedback condition presented 2 times. They were instructed to blast a loud noise to each peer after receiving feedback, by pressing the button when the volume bar appeared on screen. Instructions were given to always press the button, press the button as soon as the volume screen appeared, and to press the button as long as they wanted the noise to be blasted to their peers. Participants were exposed to the noise blast twice with the instructions; once with stepwise build-up and once at maximum intensity. During the task, the noise blast would not be heard by the participants to prevent that they would be punishing themselves. During the actual task, the SNAT consisted 3 experimental blocks of 60 trials, with each feedback condition (positive, neutral or negative) presented in 20 trials in semi-randomized order. Each peer picture was randomly assigned to a feedback condition, to ensure equal gender proportions. Each trail started with a fixation screen (500 ms), the feedback condition (2500 ms), then another fixation screen (500 ms) followed by a screen with the volume bar. The volume bar started to fill up as soon as the participant pressed the button, with a colored block appearing every 350 ms, as long as the button was pressed for. The screen with the volume bar was presented for 5000 ms, whether the volume bar was filled to maximum intensity or not. Following the volume bar screen, another fixation screen was presented (0-11550 ms), before start of the next trial (Figure 4). The SNAT was executed in a MRI scanner. MRI data was gathered for the L-CID longitudinal study. This study however only focusses on the behavioral data. In depth details on MRI data gathering can be found in Achterberg & Van der Meulen, 2019.
Figure 4. Design of the Social Network Aggression Task.
a) The three feedback conditions. The green thumbs up represents positive feedback. The grey circle represents neutral feedback. The red thumbs down represents negative feedback. (Adapted from Achterberg et al., 2016). b) Representation of the volume bar with buildup intensity of the noise blast. (Adapted from Achterberg et al., 2016). c) Representation of one trial of the SNAT, with duration of the different screens. (Adapted from Achterberg et al., 2017).
2.3.3 | SNAT stories
The SNAT stories is a variation on the SNAT. Participants were asked to rate how they felt about four characters: 1) the primary parent (i.e. their mother or father), 2) their twin brother/sister, 3) their best friend, and 4) an unknown same aged peer. Participants could rate the characters on a 6 point scale, from “not very nice” to “very nice”. During the task, the participants had to read different stories on a social situation between these characters and themselves. Participants were instructed to imagine the stories as if they were real. The story could be about the character intentionally rejecting the participant (“Ellie tells a joke to other children. She has to laugh a lot. You ask her what the joke is, but she doesn’t want to tell you.”), or unintentionally rejecting the participant (i.e. the character had a solid reason to deny the participant) (“Ellie is crafting and painting. You want to use her green paint. Ellie explains that the green paint has run out. So you cannot use it.”). After reading the story, the participant was instructed to blast a noise to the character by pressing a button when a volume bar appeared. The longer the button was pressed, the more intense the noise blast would be. Prior to the task, the participants were familiarized with the sound of the noise blast twice, once with stepwise build-up and once at maximum intensity. Just as with the original SNAT, the noise blast would not be heard by the participants during the task, to prevent them from punishing themselves. The SNAT stories consisted of one practice trial and one experimental block of eight trials. Each trial consisted of a fixation point (500 ms), followed by a story on one of the characters. After finishing reading the story, participants could press a button to go to the next screen, which was a screen with the volume bar (2000 ms). The volume bar started to fill up with colored blocks as soon as, and for as long as the button was pressed. After the volume bar, another fixation screen (500 ms) appeared, before start of the following trial (Figure 5). The four different characters were divided over the experimental trials. Each characters was presented in two of the eight trials. With one intentional rejection and one unintentional rejection story for each character. For the purpose of this study, we used the noise blast duration after the intentional rejection of the unknown peer, as this is most comparable to the SNAT measure of wave 1 and 3 (see section 2.3.2.)
Figure 5. Design of the SNAT stories.
Representation of one trial of the SNAT stories with duration per section of the trial. Original text of the trials is written in Dutch, translated to English for this thesis.
2.4 | Statistical analysis 2.4.1 | Stop-signal
A one-way repeated measures ANOVA (rmANOVA) on the SSRT was used to calculate whether there is an effect of development on the SSRT. The SSRT was the dependent variable and the four different time points the independent variable. Beforehand, the raw data was transformed to Z-scores to detect outliers. Outliers with a Z-score outside the range of 99.9% of the Z-distribution (<-3.29 or >(<-3.29) were winsorized (Tabachnick & Fidell, 2013). To test for the assumptions of the rmANOVA, a Shapiro-Wilk test of normality and Mauchly's test of sphericity were applied. A post-hoc pairwise t-test with bonferroni correction was used to determine the statistical significance of the difference between each of the four time points.
2.4.2 | SNAT and SNAT stories
A rmANOVA was applied for the SNAT at time points one and three with the three conditions (i.e. positive, negative and neutral). This was done to test whether participants understood the task and the negative feedback condition was thus a valid task for the examination of hot behavioral control. A difference in noise blast duration between the negative control and the neutral and positive feedback condition is necessary to conclude that the negative condition of the SNAT indeed measures hot behavioral control. For the SNAT stories, a four (character) by two (unintentional vs intentional) ANOVA was used to determine whether intentional rejection resulted in a longer noise blast duration than unintentional rejection.
To test for an effect of development on hot behavioral control (i.e., the duration of the noise blast), a one-way rmANOVA was used with the noise blast duration as the dependent variable and the time points as the independent variable. For the dependent variable, the noise blast duration of the negative feedback condition from the SNAT was used, and from the SNAT stories, the noise blast duration of the intentional rejection condition of the unknown character was used, since this condition is most comparable to the SNAT. A Shapiro-Wilk test of normality and the Mauchly’s test of sphericity were used to test the assumptions of the one-way rmANOVA. A poshoc pairwise t-test with bonferroni correction was applied as a pairwise comparison, to determine statistical significance of the difference between each of the four waves.
3 | Results 3.1 | Stop-signal
To detect outliers, the raw data for each time point was transformed to scores. Outliers with a Z-score outside the range of <-3.29 or >3.29 were winsorized (Tabachnick & Fidell, 2013; T1: 2 outliers, T2: 1 outliers, T3: 1 outliers, T4: 2 outliers). Missing data due to dropout or data with a percentage correct on NOGO trials < 30 were removed. The amount of missing data, outliers and the range of Z-scores for each time point is displayed in table 2.
To determine the effect of time on the SSRT (as an indication of cool control), a one-way rmANOVA was performed. The Shapiro-Wilk revealed that the assumption of normality was met for T1 (p < 0.001) and T2 (p = 0.047), and that it was violated for T3 (p = 0.159) and T4 (p = 0.804). However, since the Central Limit Theorem (Kwak & Kim, 2007) states normality can be ignored in samples larger than 30, a parametric rmANOVA was nevertheless performed. The assumption of sphericity was met (p = 0.002). The one-way rmANOVA showed a significant effect of time on the SSRT (F(3, 540) = 59.582, p < 0.001). A post-hoc pairwise t-test with bonferroni correction revealed a significant difference between the SSRT of T1 compared to T2 (p < 0.001), T3 (p < 0.001) and T4 (p < 0.001). Similarly, it revealed a significant difference between the SSRT of T2 and T3 (p = 0.042) between T2 and T4 (p = 0.005). On the other hand, the difference between the SSRT of T3 and T4 did not
significantly differ (p = 1.000), indicating that the decrease in SSRT levels off between T3 and T3. Results of the one-way rmANOVA are displayed in table 3 and figure 6.
Figure 6 shows a decrease in SSRT over the four time points. A larger decrease in SSRT is displayed between the first and second time point than between the second and third, and third and fourth time point.
Figure 6. Stop-Signal Reaction Time (SSRT) for each time point.
The mean SSRT in ms for each time point. Error bars display the 95% confidence interval. The y-axis shows the SSRT, the x-axis the four time points. * significant difference between time points of p < 0.05.
Table 3.
Results of the one-way rmANOVA on the SSRT in ms for each time point.
Time point Mean SSRT Std. error
95% Confidence Interval Lower bound Upper bound
1 330.286 5.945 318.556 342.016
2 279.414 4.213 271.101 287.726
3 265.564 4.307 257.065 274.064
4 260.702 4.637 251.551 269.852
Table 2.
Minimum and maximum Z-scores, outliers and missing or removed data for each time point.
Time point Min. Z-score Max. Z-score Winsorized outliers (N) Missing or removed (N)
1 -3.20761 5.45861 2 6
2 -3.10751 3.48563 1 21
3 -2.84530 3.48563 1 30
3.2 | SNAT and SNAT stories
A one-way rmANOVA was performed to test the effect of condition (i.e., positive, neutral, negative) on the noise blast duration after receiving social feedback during the SNAT at T1 and T3. At both time points, the assumption of sphericity was met (p < 0.001). A significant effect of condition on the noise blast duration was found with a rmANOVA for both T1 (F(2, 508) = 230.032, p < 0.001) and T3 (F(2, 452) = 759.379, p < 0.001). A post-hoc pairwise t-test with bonferroni correction revealed a significant difference between the noise blast duration after positive feedback compared to negative (p < 0.001) and neutral feedback (p < 0.001). As well as a significant difference between the noise blast duration after negative feedback and neutral feedback (p < 0.001). Table 4 and Figure 7a, figure 7c show the results of the one-way rmANOVA on the different condition of the SNAT.
To test for the effect of condition with different characters in the SNAT stories, a 4x2 rmANOVA (two conditions: intentional and unintentional; for every character: primary parent, friend, twin and unknown peer) was performed. The assumption of sphericity was violated for the variable characters of T2 (p = 0.079) and for the interaction between the conditions and the characters of T2 (p = 0.828) For the interaction of T4 the assumption was met with a p-value of 0.003, as well as for the variable characters of T4 (p = 0.010). For both time points, a one-way rmANOVA with Greenhouse-Geisser correction revealed a significant effect of condition (T2: F(1.000, 255.000) = 197.302, p < 0.001; T4: F(1.000, 196.000) = 277.808, p < 0.001) as well as for character (T2: F(2.923, 745.423) = 21.230, p < 0.001; T4: F(2.854, 559.401) = 19.823, p < 0.001) and the interaction between condition and character (T2: F(2.983, 760.750) = 6.622, p < 0.001, T4: F(2.824, 553.525) = 19.216, p < 0.001) on the noise blast duration. Figure 7b and figure 7d show the results of the one-way rmANOVA on the different condition and characters of the SNAT stories.
To test for the effect of development on the noise blast duration (as an indication of hot control), a one-way rmANOVA was performed. The assumption of sphericity was met (p < 0.001). The one-way rmANOVA revealed a significant effect of development on the noise blast duration with a p-value of <0.001 (F(3) = 28.390). A post-hoc pairwise t-test with bonferroni correction revealed a significant difference between the noise blast duration of T1 compared to T2 (p < 0.001) and T4 (p < 0.001), however, not compared to T3 (p = 1.000). It revealed a significant difference between the noise blast duration of T2 compared to T3 (p < 0.001), but not compared to T4 (p = 0.247). And lastly, a significant difference between the noise blast duration of T3 and T4 was revealed (p < 0.001). Results of the one-way rmANOVA on the noise blast duration over the waves are displayed in table 5 and figure 8.
Based on the results of the rmANOVA, a Pearson correlation was conducted to determine whether data from the SNAT and SNAT stories could be compared. Low correlations were found for T1 compared to T2 (r = 0.09, p = 0.154), and compared to T4 (r = -0.008, p = 0.906), as well as for T2 compared to T3 (r = 0.071, p = 0.285), and for T3 compared to T4 (r = 0.115, p = 0.109). Significant correlations were found for T1 compared to T3 (r = 0.167, p = 0.012), and for T2 compared to T4 (r = 0.213, p = 0.003). Results of the Pearson correlation between the time points are displayed in table 6.
Figure 7. Noise blast duration per condition for the SNAT, and per character per condition for the SNAT stories.
a) The noise blast duration in ms for every condition for T1, b) the noise blast duration in ms for every condition and character for T2, c) the noise blast duration in ms for every condition for T3, and d) the noise blast duration in ms for every condition and character for T2. Error bars display the 95% confidence interval. * significant difference between condition/character of p < 0.05.
Table 4.
Noise blast duration in ms of the SNAT for timepoints 1 and 3.
Condition Mean noise blast duration Std. error
95% Confidence Interval Lower bound Upper bound
T1 Positive 1486.092 54.025 1379.698 1592.486 Neutral 1911.731 40.423 1832.123 1991.338 Negative 2726.816 43.917 2640.328 2813.304 T3 Positive 929.705 36.213 858.346 1001.064 Neutral 1664.636 37.178 1591.376 1737.896 Negative 2714.509 46.951 2621.990 2807.028
Figure 8. Mean noise blast duration in ms for each of the time points.
Error bars show the 95% confidence interval. * significant difference of the noise blast duration between the time points of p < 0.05.
Table 5.
Noise blast duration in ms for every time point.
Time point Mean noise blast duration Std. error
95% Confidence Interval Lower bound Upper bound
1 2699.819 51.580 2598.090 2801.548
2 2255.585 77.714 2102.312 2408.857
3 2721.976 51.815 2619.782 2824.169
4 2054.272 77.509 1901.403 2207.141
Table 6.
Results of the Pearson correlation between each of the four time points.
1, SNAT 2, SNAT stories 3, SNAT 4, SNAT stories
1, SNAT Correlation 1.000 0.09 0.167 -0.008
p-value 0.154 0.012 0.906
2, SNAT stories Correlation 0.09 1.000 0.071 0.213
p-value 0.154 0.285 0.003
3, SNAT Correlation 0.167 0.071 1.000 0.115
p-value 0.012 0.285 0.109
4, SNAT stories Correlation -0.008 0.213 0.115 1.000
4 | Discussion
The aim of this study was to determine the development of both cool and hot behavioral control between the ages 7-9 to 10-12 years. The development of cool behavioral control was investigated using the Stop-signal task. The development of hot behavioral control with the Social Network Aggression Task, developed by Achterberg and colleagues (Achterberg et al., 2016), and a variation on the SNAT, the SNAT stories. Results from this study could help narrow the knowledge gap on the developmental trajectories of hot and cool behavioral control and how those might be different. 4.1 | The development of cool behavioral control
The development of cool behavioral control was investigated with the Stop-signal task. It was used in all four time points of data gathering of this study. A significant decrease of SSRT between the ages 7-9 years and 10-12 years was found. Moreover, a significant decrease of SSRT between the first time point (ages 7-9 years) and the second (8-9 years), and between the second and the third (9-10 years) was found. From this, it can be concluded that there is a significant increase in cool behavioral control between the ages 7 to 10 years. The decrease of SSRT between the third and fourth time point (10-12 years) was however not significant. This suggests that the development of cool behavioral control levels off at the age of 10. These results are not completely in line with the expected decrease in developmental speed of cool behavioral control from the age of 8 years on, based on previous research of Lewis et al. (2017). It does however confirm a decrease in developmental speed before adolescence, only later than expected from previous research.
The pattern of development of cool behavioral control is conform the hypothesis stated in section 1. As expected, an increase of cool behavioral control was found between the ages 7-9 to 10-12, with an exponential decrease of developmental growth between the time points. Not only did the results of this study match the results of previous research, it provides evidence of within-subject developmental changes, while most previous research suggested changes over time based on cross sectional research (Carver, Livesey & Charles, 2001; Ridderinkhof, Band & Logan, 1999).
The hypothesis also stated that the development of cool behavioral control was expected to continue until early adulthood. Since the decrease of SSRT, and therefore the increase of cool behavioral control, is not significant between the third and last time point, it cannot be concluded with certainty that the development is ongoing. However, there is an increase in cool behavioral control, though not significant, therefore it could be suggested that the development does continue but at a rate that is too slow to significantly measure within one year. This could be explained by the exponential decrease of the increase. Before the stationary phase, where the development has stopped, the rate of development is very low. To confirm or reject the hypothesis that the development of cool behavioral control is ongoing between the ages 9-10 and 10-12, future studies should include more time points to really capture within-person development from childhood to adulthood.
4.2 | The development of hot behavioral control
The development of hot behavioral control was examined using the SNAT and SNAT stories. The SNAT was used during the first and third time point of data gathering, the SNAT stories for the second and the fourth. A significant increase of noise blast duration for the negative compared to the neutral and the positive condition was found with the SNAT. A significant increase of noise blast duration for the intentional rejection compared to the unintentional rejection condition was found with the SNAT stories. From this, the conclusion can be drawn that both the SNAT and the SNAT stories are valid tasks to measure hot behavioral control. The noise blast duration significantly decreased between the first (7-9 years) and the last time point (10-12 years). However, results from the first time point and last time point were gathered with different tasks (i.e., the SNAT and SNAT
stories) and between the time points within these tasks (T1 and T3 for the SNAT; T2 and T4 for the SNAT stories) the results did not significantly differ. Besides, between the SNAT and SNAT stories, the Pearson correlation revealed non-significant correlations. From this, it can be concluded that results from the SNAT and the SNAT stories cannot be compared to determine the development of hot behavioral control. So, from the results comparing the four time points, a strong conclusion on the development of hot behavioral control cannot be drawn.
As shown in figure 2, a developmental pattern of exponential increase of hot behavioral control was expected. Despite the significant decrease in noise blast duration between the first and last time point, and therefore an increase in hot behavioral control, the hypothesis can neither be confirmed nor rejected by the results of the SNAT and SNAT stories over the four time points due to the fact that the results of the SNAT and SNAT stories are not suitable to use for one variable. Besides that, within the two different tasks, the results were also not significant. However, even though the results within the different tasks did not show a significant decrease of noise blast duration, an exponential increase of hot behavioral control could still be possible. With an exponential increase of hot behavioral control, the changes between the first time points could be too little to be significant. Seeing that the correlations between T1 and T3, measured with the SNAT, and between T2 and T4, measured with the SNAT stories, were significant. The non-significant results could be an indication that the development of hot behavioral control only noticeably increases after middle childhood. This is in line with the imbalance theory of Casey et al. (2008), explained in section 1. Their theory on the imbalance between a heightened responsiveness and the control over the responsiveness, which is noticeable between childhood and adulthood, could explain the non-significant development of hot behavioral control during middle childhood. According to the imbalance theory, the control over the heightened responsiveness (i.e., hot behavioral control), noticeably develops during the end of the adolescence. In order to draw a solid conclusion on the pattern of development of hot behavioral control, more time points for both the SNAT and the SNAT stories should be included.
The non-comparability of the tasks could be explained by the differences between the SNAT and SNAT stories. First of all, the environment in which the tasks are being conducted are different. The SNAT is conducted in a MRI-scanner in a medical lab, while the SNAT stories is conducted at home on a laptop. Both time points from the SNAT have significantly higher noise blast duration scores than the time points from the SNAT stories. Generally, the participants are more comfortable at home than at the medical lab. This might have affected the degree in which the participants reacted to negative feedback. Second, both the SNAT and the SNAT stories provide fictional situations to which the participants have to react. However, prior to the SNAT, participants are told the characters have seen their profiles and have given feedback. With the SNAT stories, the participants receive fictional stories which they have to imagine and react to. Even though both situations are fictional, the SNAT stories could be perceived as more fictional than the SNAT, since participants are explicitly told the stories are fictional. This could also have affected the reaction of the participants. Lastly, the SNAT has three conditions, a negative, a neutral, and a positive feedback condition, while the SNAT stories has two conditions, the intentional and unintentional rejection condition. It could be that participants perceived the negative feedback condition as more negative compared to the positive feedback condition, than they did with the intentional rejection condition compared to the unintentional condition. Participants could have scored higher on the noise blast duration in the SNAT, due to a bigger contrast between the positive and the negative condition in the SNAT than in the SNAT stories.
4.3 | Further research
The need for a task that measures hot behavioral control in a social setting, that can be adjusted to age and can be used for several consecutive years without the occurrence of a learning effect, has become apparent. The SNAT, as well as the variation on the original, the SNAT stories, are valid tasks for measuring hot behavioral control in a social setting. Further research should determine whether a learning effect occurs when the SNAT is used for several consecutive years. A third time point with the use of the SNAT (as is the case in the L-CID study) could bring clarification on the development of hot behavioral control.
As discussed in section 1 and 4.2, the development of hot behavioral control was expected to follow a different pattern as that of cool behavioral control, with an exponential increase of behavioral control. This could not be investigated with the results of the current study. Further research could compare the developmental patterns of both hot and cool behavioral control when both would be examined on a longitudinal level. Different from the current study, hot behavioral control should be examined using one task.
4.4 | Conclusion
Using the Stop-signal task to measure cool behavioral control, this study provided evidence for the development of cool behavioral control between the ages 7 to 10 years. Further research should provide insight on the further course of the development. The results of the current study contribute to the knowledge on the development of cool behavioral control. A conclusion on the development of hot behavioral control could not be drawn with the results of the SNAT and SNAT stories. However, evidence that both the SNAT and SNAT stories are valid tasks to measure hot behavioral control was provided. Further research should look into the use of the SNAT or SNAT stories as separate tasks to determine the development of hot behavioral control. Finally, with the results of this study, another step in the understanding of the development of behavioral control and the differences between aspects of behavioral control has been made.
5 | References
Achterberg, M. & Van der Meulen, M. (2019). Genetic and environmental influences on MRI scan quantity and quality. Developmental Cognitive Neuroscience, 38. https://doi.org/10.1016/j.dcn.2019.100667
Achterberg, M., van Duijvenvoorde, A. C. K., Bakermans-Kranenburg, M. J. & Crone, E. A. (2016). Control your anger! The neural basis of aggression regulation in response to negative social feedback. Social Cognitive and Affective Neuroscience, 2016, 712–720. DOI: 10.1093/scan/nsv154 Achterberg, M., van Duijvenvoorde, A. C. K., van der Meulen, M., Euser S., Bakermans-Kranenburg
M. J. & Crone E. A. (2017). The neural and behavioral correlates of social evaluation in childhood, Developmental Cognitive Neuroscience, 24: 107-117.
Anderson, P. (2002). Assessment and development of executive function (EF) during childhood. Child Neuropsychology, 8, 71–82. DOI: 10.1076/chin.8.2.71.8724
Carver, A. C., Livesey, D. J., Charles, M. (2001). Age related changes in inhibitory control as measured by stop signal task performance. International Journal of Neuroscience, 107 (1-2): 43 - 61.
Casey, B. J., Jones, R. M., Hare, T. A. (2008). The Adolescent Brain. Annals of the New York Academy of Sciences, 1124: 111-126. doi: 10.1196/annals.1440.010
De Water, E., Cillessen, A. H. N. & Scheres, A. (2014). Distinct Age-Related Differences in Temporal Discounting and Risk Taking in Adolescents and Young Adults. Child development, 85 (5): 1881-1897. https://doi.org/10.1111/cdev.12245
Diamond, A. (2013). 'Executive functions', Annual Review of Psychology, 64: 135-168. DOI:10.1146/annurev-psych-113011-143750
Euser, S., Bakermans-Kranenburg, M. J., van den Bulk, B. G., Linting, M., Damsteegt, R. C., Vrijhof, C. I., van Wijk, I. C., Crone, E. A. & van IJzendoorn, M. H. (2016). Efficacy of the Video-Feedback Interaction to promote Positive Parenting and Sensitive Discipline in Twin Families (VIPP-Twins): Study protocol for a randomized controlled trial. BCM Psychology, 4: 33. DOI: 10.1186/s40359-016-0139-y
Kerr, A. & Zelazo, P. H. (2004). Development of “hot” executive function: The children’s gambling task. Brain and Cognition, 55 (1): 148-157. https://doi.org/10.1016/S0278-2626(03)00275-6 Kwak, S. G., & Kim, J. H. (2007). Central limit theorem: the cornerstone of modern statistics. Korean
Journal of Anesthesiology, 70 (2): 44- 156. doi: 10.4097/kjae.2017.70.2.144
Leary, M. R., Gallagher, B., Fors, E., Buttermore, N., Baldwin, E., Kennedy, K. & Mills, A. (2003a). The Invalidity of Disclaimers about the Effects of Social Feedback on Self-Esteem. Personality and Social Psychology Bulletin, 29 (5), 623–636. https://doi.org/10.1177/0146167203029005007 Leary, M. R., Kowalski, R. M., Smith, L. & Philips, S. (2003b). Teasing, Rejection and Violence: Case
Studies of the School Shootings. Aggressive Behavior, 29 (3): 202- 214. DOI: 10.1002/ab.10061 Lewis, F. C., Reeve, R. A., Kelly, S. P. & Johnson, K. A. (2017). Evidence of substantial development
of inhibitory control and sustained attention between 6 and 8 years of age on an unpredictable Go/No-Go task. Journal of Experimental Child Psychology, 157: 66-80. doi: 10.1016/j.jecp.2016.12.008.
Ridderinkhof, K. R., Band, G. P. H., Logan, G. D. (1999). A study of adaptive behavior: effects of age and irrelevant information on the ability to inhibit one's actions. Acta Psychologica, 101 (2–3): 315-337. https://doi.org/10.1016/S0001-6918(99)00010-4
Scheres, A., Dijkstra, M., Ainslie, E., Balkan, J., Reynolds, B., Sonuga-Barke, E. & Castellanos, F. X. (2006). Temporal and probabilistic discounting of rewards in children and adolescents: Effects of age and ADHD symptoms. Neuropsychologia, 44 (11): 2092-2103. https://doi.org/10.1016/j.neuropsychologia.2005.10.012
Silvers, J. A., McRae, K., Gabriele, J. D. E, Gross, J. J., Remy, K. A., Ochsner, K. N. (2012). Age-Related Differences in Emotional Reactivity, Regulation and Rejection Sensitivity in Adolescence. Emotion, 12 (6): 1235–1247. DOI: 10.1037/a0028297
Strommen, E. A. (1973). Verbal Self-Regulation in Children’s Game: Impulsive Errors on “Simon Says”. Child Development, 44 (4): 849-853. DOI: 10.2307/1127737
Tabachnick, B. G. & Fidell, L. S. (2013). Using Multivariate statistics (6th edition). Boston, Pearson. Tillman, C. M., Thorell, L. B., Brocki, K. C. & Bohlin G. (2007). Motor Response Inhibition and
Execution in the Stop-Signal Task: Development and Relation to ADHD Behaviors. Child neuropsychology, 14 (1): 42-59. https://doi.org/10.1080/09297040701249020
Twenge, J. M., Baumeister, R., Tice, D. & Stucke, T. S. (2001). If you can’t join them, beat them: Effects of social exclusion on aggressive behavior. Journal of Personality and Social Psychology, 81 (6): 1058-69. DOI: 10.1037//0022-3514.81.6.1058
Williams, B. R., Ponesse, J. S., Schachar, R. J., Logan, G. D., & Tannock, R. (1999). Development of inhibitory control across the life span. Developmental Psychology, 35 (1): 205–213. https://doi.org/10.1037/0012-1649.35.1.205
Zelazo, P. D. & Carlson, S. M. (2012). Hot and Cool Executive Function in Childhood and Adolescence: Development and Plasticity. Child Development Perspectives, 6 (4): 354-360. https://doi.org/10.1111/j.1750-8606.2012.00246.x
