Background: Children have a profound ability to learn from the environment through information gathering and updating. Unfortunately, adversity that co-occurs with brain development can severely impact learning processes (Pechtel & Pizzagalli, 2011). To date, studies of caregiving related early adversity (crEA) have focused primarily on higher-level learning (e.g. reward learning and fear conditioning), but have not yet focused on basic forms of learning. In fact, basic forms of learning may underlie deficits in higher-level learning and complex behaviors (e.g., executive function (Pechtel & Pizzagalli, 2011)) that are often impaired following crEA-exposure. One such form of learning, implicit sequence learning, emerges in infancy (Meulemans et al., 1998; Saffran et al., 1996) and is central to multiple domains including language (Saffran et al., 1996; Ullman, 2001), cognitive and social skills (Lieberman, 2000) and executive functions (Wassenberg et al., 2017), central to future academic and social success. These findings merit investigation of basic learning in crEA-exposed children.
Motor sequence learning (MSL) involves extracting implicit patterning information and, at the level of the brain, involves the coordination of multiple neural regions including frontal cortex (prefrontal cortex, motor cortex, pre-supplemental motor area), parietal cortex, and basal ganglia (Hikosaka et al., 2002; Mattar & Bassett, 2016). During the developmental period of 6-12 years, MSL and associated neural regions undergo significant maturation (Chai et al., 2017; Meulemans et al., 1998; Wassenberg et al., 2017). In this study, I will investigate how crEA-exposure (e.g., foster care, international adoption, caregiver separation) impacts implicit motor sequence learning in 6–12-year-old children.
Research Question: The main hypothesis of this paper states that implicit motor sequence learning in school-age children following crEA-exposure is associated with adversity-related deficits in implicit learning, that underlies higher level cognitive processes, such as executive function.
Analyses & Results (in R):
No crEA vs. comparison group difference in sequence learning. A repeated measures ANOVA analysis showed a main effect of block but no significant effect of group or GroupXBlock interaction. Results suggest the comparison and the crEA group both learned the sequence, as indicated by comparable response times (RTs) on the task, with no significant group differences. Block 1, Block 2 and Block 4 are repeated sequence blocks, and Block 3 is a random sequence. Plotted are group average median RTs per block and standard error bars. *** indicates significance at p < .001 in post-hoc tests.
Explicit learning. A. Recall test. Correct responses. Groups show similar average correct responses on the recall test, with an average of .51 for both groups. B. Recall test. Streak length. Groups show similar average streak length, with an average streak length of 3.27 for the comparison group and 3.29 for the crEA group. C. Recognition test. There was no significant difference in average correct responses on the recognition test, with an average of .43 for the comparison group and .36 for the crEA group. Bars represent group average responses with standard error bars, and points (jittered) represent individual participants’ average responses for A. number correct score, out of 9 questions, B. highest streak length score range 1-9, and C. number of correct score, out of 2 questions.
Experiential knowledge questions. A. Question 1 asked participants if they thought they got faster at catching minions and B. Question 2 asked participants if they thought they could ever tell where the next minion would appear. The majority of children believed they learned and could recognize, at least in part, the pattern of the sequence. There were no group differences in frequency of responses in either question.
Gender differences across blocks. Post-hoc analyses of the GenderXBlock interaction in the primary implicit motor sequence learning model revealed that the interaction was driven by differences in learning, as indicated by significant RT differences across Block 1 compared to Block 3 (F(1, 226) = 8.19, p = .005), Block 2 compared to Block 4 (F(1, 226) = 5.56, p = .019), and trending differences in Block 3 compared to Block 4 (F(1, 226) = 5.26, p = .022). Paired t-tests found that these gender differences were driven by significant differences in Block 1 compared to Block 4 for males (t(109) = 4.55, p < .001), who showed significantly faster average RTs in Block 4, and significant differences in Block 2 and Block 4 for females (t(119) = -2.04, p = .043), who showed slower average RTs in Block 4. Overall, boys demonstrated consistent speeding up across sequence Blocks (Block 1, Block 2, and Block 4) while girls did not show speeding up on the last sequence Block. Plotted are group average median RTs per block for each gender and standard error bars. *** indicates significance at p < .001, ** indicates significance at p < 0.01, and * indicates significance at p < .05 in post-hoc tests.
Mixed effects multi-level model. A BlockXGroup interaction was tested, controlling for age and gender, allowing for individual variability within subjects per block using the following equation, in brms syntax:
RT_correct = Intercept + Group (comps versus crEAs) + Block + Sex + Age + GroupXBlock + (Block | Participant)
Mixed-effects model outputs indicating no group difference in sequence learning. A mixed-effects model approach allowing for within subject variability across blocks, there was a possible slight main effect of Block 1sequence and no group differences across the subsequent blocks, Block 2sequence, Block 3random, or Block 4sequence).Overall, the model suggests that crEA were slightly slower than the comparison group to learn the sequence but do not show differences in learning. Both groups learn the sequence, as indicated by a slower RT in the random block (Block 3) compared to faster learning blocks. Plotted are model predicted means (error bars are standard errors) for each group RT across blocks. The comparison group is in black (left) and the crEA group is in blue (right).
Task Design/Methods (designed using PsychoPy/Python):
Task design. A. Serial reaction time (SRT) Task Participants were told to “catch” the minions by touching the location of the minion as fast as possible. The first, second, and fourth block contained a fixed sequence of locations (repeated sequence depicted in C. (1)), and the third block was a random order. Following the SRT task, participants were asked a series of questions to assess explicit knowledge of the task structure. B. Explicit recall test. Participants were asked “Where do you think the next minion will be?” 9 times to complete one full cycle of the sequence. C. Explicit recognition test. Participants were asked twice, “Which pattern did you press? 1, 2, or 3” with only one correct response.
Findings and Implications:
Basic learning (of motor sequences) is not impacted or altered by early life caregiving adversity
Higher-order functioning, such as executive functions (impulse and cognitive control), is impacted by early life stress
Motor sequence learning does not underly alterations in higher-order functions followed by early life stress.
Interventions with at-risk children should target higher-order functions and not basic learning
Comments