Children’s Allocation of Study Time during the Solution of Raven’s Progressive Matrices

Abstract

The acuity of reasoning on Raven’s Progressive Matrices is strongly influenced by strategic determinants. Building on metamemory studies that highlight the influence of study-time allocation on memory development, we investigated children’s allocation of study time while solving these matrices. A total of 170 children aged 6–12 years completed a computerized short-form version of the standard matrices featuring items selected to represent a broad range of difficulties. Beyond analyzing changes in mean latencies and performances with age, we used generalized additive mixed models to explore within-participant variability in response times as a function of both item complexity and overall individual efficiency. Results revealed that individual differences in performances were significantly associated with children’s adaptive modulation of response times. Mediation analysis further indicated that response-time modulation contributed to age-related changes in performance. Taking account of study-time allocation in reasoning tasks may open up new avenues for the study of reasoning development and the assessment of intellectual functioning.

Intraindividual strategy shifts in Raven’s matrices, and their dependence on working memory capacity and need for cognition.

“…Such a conclusion would be at odds with the extensive literature showing that individuals flexibly adapt their strategies as best as possible to face the current problem (Lemaire & Siegler, 1995;Siegler & Shipley, 1995). However, this apparent ambiguity is resolved when considering performance beyond just accuracy on the task: the present results suggest that maintaining constructive matching over the most difficult items is associated with disproportionately increased response times (see Perret & Dauvier, 2018, for a similar example), and, as illustrated by the interaction with WMC and as hypothesized by the literature, with a high working memory load (e.g., Bethell-Fox et al, 1984;Mulholland et al, 1980;. In this sense, the shift toward response elimination is actually adaptative: rather than just accuracy, it optimizes the tradeoff between invested effort and likelihood of success.…”

“…NFC can be viewed as the motivation to expand effort to perform costly cognitive processing (e.g., Sandra & Otto, 2018). While it is straightforward to understand why this could lead to higher performance in intelligence tasks (Perret & Dauvier, 2018; Raven, Raven, & Court, 1998), this definition also suggests a relation with strategy use. Because constructive matching is viewed as more costly, participants with a high NFC may be more likely to engage in this strategy.…”

“…Because constructive matching use is moderately associated with longer time on the problem (Gonthier & Thomassin, 2015; Mitchum & Kelley, 2010; Vigneau et al, 2006; see also Foorman, Sadowski, & Basen, 1985; Partchev & De Boeck, 2012), response times were also investigated. Response times have the advantage of constituting an objective behavioral measure, and intraindividual changes in response times have been found to offer important insight into participants’ engagement and invested effort (Goldhammer, Naumann, & Greiff, 2015; Perret & Dauvier, 2018). Critically, participants performed the APM with no time limit to allow for unconstrained strategy use (see Thomassin et al, 2015).…”

“…Both features make GAMMs particularly suited to the analysis of intraindividual variability. GAMMs have recently begun emerging in psychological studies of intraindividual variability (for an example in the field of intelligence, see Perret & Dauvier, 2018; for other examples, see Baayen, Vasishth, Kliegl, & Bates, 2017; Le Vigouroux, Pavani, Dauvier, Kop, & Congard, 2017; McKeown & Sneddon, 2014). In the present study, GAMMs were used to systematically model the timecourse of accuracy, response time, constructive matching, and response elimination use throughout Raven’s APM; as allowed by this type of analysis, the shape of these timecourses—linear or not—was directly estimated from the data.…”

“…Individual items thus constitute a meaningful level of analysis. In this light, an intelligence task can be seen not as an indivisible test yielding a single score, but as a sequence of problems, whose succession induces positional effects such as modulation of exerted effort (Perret & Dauvier, 2018) and complex learning of one’s successes and failures (Ren, Wang, Altmeyer, & Schweizer, 2014). One possibility, and the focus of the present study, is that successive problems also elicit intraindividual changes of strategy (Schunn & Reder, 2001).…”