Understanding the What and When of Analogical Reasoning Across Analogy Formats: An Eye‐Tracking and Machine Learning Approach

Abstract

Starting with the hypothesis that analogical reasoning consists of a search of semantic space, we used eye-tracking to study the time course of information integration in adults in various formats of analogies. The two main questions we asked were whether adults would follow the same search strategies for different types of analogical problems and levels of complexity and how they would adapt their search to the difficulty of the task. We compared these results to predictions from the literature. Machine learning techniques, in particular support vector machines (SVMs), processed the data to find out which sets of transitions best predicted the output of a trial (error or correct) or the type of analogy (simple or complex). Results revealed common search patterns, but with local adaptations to the specifics of each type of problem, both in terms of looking-time durations and the number and types of saccades. In general, participants organized their search around source-domain relations that they generalized to the target domain. However, somewhat surprisingly, over the course of the entire trial, their search included, not only semantically related distractors, but also unrelated distractors, depending on the difficulty of the trial. An SVM analysis revealed which types of transitions are able to discriminate between analogy tasks. We discuss these results in light of existing models of analogical reasoning.

Cognitive segmentation and fluid reasoning in childhood

“…As in Duncan et al's (2017) study, the majority of errors in both Combined and Separated conditions were "wrong alternatives," that is, drawing the alternative feature that was presented in the matrix, but was not the correct response feature. This might suggest that the children were easily confused when selecting which features to attend and select from cells A, B, or C, and is consistent with the broader literature on children's problem-solving skills, suggesting that children do not systematically orient and organise their searches across the cells in matrix reasoning problems and analogical reasoning more generally (Chen et al, 2016;Glady et al, 2012Glady et al, , 2017Starr et al, 2018;Thibaut & French, 2011;Vendetti et al, 2017). While the Separated format reduced errors of all types, and did not induce a qualitatively different pattern of errors, it appeared particularly beneficial in reducing the most common "wrong alternative" errors.…”

The Development of Flexible Problem Solving: An Integrative Approach

“…Gaze behaviors are a promising source of information on cognitive development, including the development of flexible problem solving because, they can be gathered across ages, and do not exclude young children with poor motor control and eye-hand coordination. Looking-time measures have often been used in psychological research to investigate the early and lifespan development of cognitive capacities (e.g., Baillargeon 2004;Eckstein et al 2017;Krøjgaard et al 2020;Onishi and Baillargeon 2005;Thibaut et al 2011). Looking-time measures may be collected even in the first days of life, offering early insight into developmental changes in attention.…”

“…Increases in knowledge may also support such generalization, but it is unlikely that only 14-year-olds would have conceptual knowledge sufficient for transferring simple relations, such as "X fits in Y" across simple items that differ in colors and shapes. Eightyear-olds' disadvantage may instead result from poorer use of executive functions than 14-year-olds' (Thibaut et al 2011;Thibaut and French 2016). Thibaut and French (2016) showed that 8-year-olds focus more on the C item and less on the A:B pair than adults, suggesting that, compared to 14-year-olds, 8-year-olds may have greater difficulties in inhibiting the ultimate goal of the task (studying C to find the correct D) in favor of the currently relevant subgoals (studying the relationship between A and B; drawing the relationship between C and potential Ds), and switching between the ultimate goal and the subgoals (see also French et al 2017 for a modelling approach).…”

“…On the one hand, knowledge about the early and lifespan development of simple and complex transfers was accumulated decade after decade, so insights into such transfers could be presented decadewise. On the other hand, complex transfers employed in analogical reasoning were already considered in antiquity (Goswami 1991;Pellegrino 1985) but started to draw researchers' attention only in 1980s and would be systematically studied even later, in 2000s (e.g., Richland et al 2006;Thibaut et al 2011;. In principle, the historical approach, outlining how philosophers' and psychologists' interest in transfers of knowledge unfolded over time, could have been adopted in this review paper.…”

Measuring the building blocks of everyday cognition: executive functions and relational reasoning

“…Relational attention may shift with a reasoner’s expertise in the relevant knowledge-base, which changes the nature of reasoners’ attention to the relational content of a task ( Chi et al, 1981 ). As knowledge increases, reasoners may shift from attention to surface features and object-level correspondences to relational correspondences (see Gentner, 1988 ; Starr et al, 2018 ; Thibaut et al, 2022 ).…”