New insights into statistical learning and chunk learning in implicit sequence acquisition

Abstract

Implicit sequence learning is ubiquitous in our daily life. However, it is unclear whether the initial acquisition of sequences results from learning to chunk items (i.e., chunk learning) or learning the underlying statistical regularities (i.e., statistical learning). By grouping responses with or without a distinct chunk or statistical structure into segments and comparing these responses, previous studies have demonstrated both chunk and statistical learning. However, few studies have considered the response sequence as a whole and examined the temporal dependency of the entire sequence, where the temporal dependencies could disclose the internal representations of chunk and statistical learning. Participants performed a serial reaction time (SRT) task under different stimulus interval conditions. We found that sequence learning reflected by reaction time (RT) rather than motor improvements represented by movement time (MT). The temporal dependency of RT and MT revealed that both RT and MT displayed recursive patterns caused by biomechanical effects of response locations and foot transitions. Chunking was noticeable only in the presence of the recurring RT or MT but vanished after the recursive component was removed, implying that chunk formation may result from biomechanical constraints rather than learning itself. In addition, we observed notable first-order autocorrelations in RT. This trial-to-trial association enhanced as learning progressed regardless of stimulus intervals, reflecting the internal cognitive representation of the first-order stimulus contingencies. Our results suggest that initial acquisition of implicit sequences may arise from first-order statistical learning rather than chunk learning.

The "Motor" in Implicit Motor Sequence Learning: A Foot-stepping Serial Reaction Time Task

“…Unlike response time and RT, MT is comparable between blocks 4 and 5 (difference = -7.73 ms ± 3.88; p = 0.072). The same RT and MT results have been reported in our other studies 3,5 . These results together suggest that sequence learning is most likely to be reflected by RT, a proxy to cognitive processing, rather than MT, which characterizes the movement itself.…”

“…In our recent studies, we extended the classic SRT task to a modified SRT task where finger pressing was replaced by foot stepping to incorporate postural control into sequence learning 3,4,5 . This modified task presents its own advantages to complement the classic SRT task.…”

Children and Adults Both Learn Motor Sequences Quickly, But Do So Differently

“…However, a subsequent study from our lab asked children and adults to perform a self-paced SRT task where the response-stimulus-interval was fixed and thus ISI was dictated by an individual’s own response speed. This subsequent study provided preliminary evidence that the online and offline process in children and adults were unlikely to be influenced by the task pacing condition ( Du, 2016 ). During the whole experiment, an accurate hit on the target mat or home position was encouraged, but not strictly required because, during the continuous stepping movement, participants, and especially children, would shift their positions slightly.…”

“…There was a significant effect of block on MT ( F 2,29 = 2.95, p < 0.05, η 2 = 0.07) but the MT differences between blocks were within 0.01 s. Post hoc tests failed to identify further differences between each pair of blocks ( Figure 4A ). In particular, MT was comparable between blocks 4 and 5, suggesting that MT did not represent sequence learning ( Du and Clark, 2016 ). We further measured the MT changes from blocks 4 to 5 for the first 50 trials and last 50 trials, respectively, and did not find significant effects of age, block phase, and their interaction ( Figure 4B ).…”

“…However, the online process may not be behaviorally expressed as the iterative mental computation after each step further shortened the amount of time available before the next stimulus appears and thus may interfere with the preparation of the succeeding step, which subsequently prolongs the following RT. To further test the task pacing effect, in a subsequent study from our lab, adults and children performed a self-paced SRT task where the ISI was determined by an individual’s own response speed ( Du, 2016 ). That is, the response-stimulus-interval was fixed at 700 ms for both children and adults and the children’s slower response speeds yielded longer ISI.…”