Strategy selection for cognitive skill acquisition depends on task demands and working memory capacity

a r t i c l e i n f o The involvement of working memory capacity (WMC) in ruled-based cognitive skill acquisition is well-established, but the duration of its involvement and its role in learning strategy selection are less certain. Participants (N = 610) learned four logic rules, their corresponding symbols, or logic gates, and the appropriate input–output combinations in three-gate circuit patterns. Participants practiced 120 repetitions of each rule (480 total gates) over the course of 10 training blocks. Memory load varied between subjects. The confluence of task demands and individual differences in WMC (N = 518) dramatically affected speed– accuracy tradeoffs and strategic use of a computerized help function. Cluster analysis revealed five distinct groups of participants based on the combination of response accuracy, latency, and help use. Some groups with moderate to high mean WMC acquired the task with predictable performance patterns. Other groups, prevalently under a memory load and with low mean WMC, failed to learn or overused help. Cognitive abilities, task constraints and strategy use each influence cognitive skill acquisition (CSA). Although these variables frequently appear in the literature, the study of their confluence is uncommon and neglected. Here we explore their interaction on learning a novel and complex logic-gate task (LGT). The LGT requires using logical functions, based on Boolean mathematics, to determine a unitary output (0 or 1) from binary input combinations (Gitomer, 1988). Four input combinations (0–0, 1–0, 0–1, 1–1) and four gate-types (AND, OR, NAND, NOR) yield 16 gate-by-input combinations. Table 1 shows the truth table of gates, rules and input–output combinations in the LG task. After an introduction to the rules, individuals practiced these gates with an option to use a " help " key to refresh their memory as needed. ACT model, for example, WM holds all declarative rules and productions until propositions are compiled, strengthened, and eventually automatized through practice and chunking (Rosenbloom & Newell, 1987; Woltz, 1988). Consistent with this assumption, individual differences in working memory capacity (WMC) and related general cognitive abilities predict skilled performance. These abilities correspond to the rate and/or accuracy of One purpose of this study is to determine the extent to which complex WM span tasks predict performance during practice. measures capture individual differences in attentional control, and attentional control is critical for complex CSA. We predict that the demand for attentional control is greatest during early learning when memory representations …