When illustrations mislead: cognitive load and the simple view of reading in scientific text learning

Scientific text learning often uses instructional illustrations to support learners' understanding of abstract concepts, processes, and causal mechanisms. However, illustrations may mislead learning when they resemble conceptual illustrations but contain incorrect mechanistic information. Drawing on multimedia learning theory, cognitive load theory, and the simple view of reading, this study examined how misleading illustrations affect learning from short scientific expository texts involving causal mechanisms and whether cognitive load and language-related abilities explain these effects.

Using a randomized three-condition between-subjects experimental design, 179 learners were assigned to conceptual-illustration, misleading-illustration, or plain-text conditions.

Compared with plain text and conceptual illustrations, misleading illustrations significantly reduced inference and overall learning performance. Mediation analysis showed that this negative effect was partly explained by increased cognitive load. Moderation analysis further indicated that reading comprehension consistently buffered the detrimental effects of misleading illustrations, whereas decoding ability did not show a comparable protective role. Item-level mixed-effects models supported the robustness of these findings.

These findings suggest that, within the short scientific expository texts examined here, mechanistic consistency and reading comprehension may represent important conditions shaping effective text-picture integration.