Sequential processing in comprehension of hierarchical graphs

Abstract

Hierarchical graphs represent the relationships between non-numerical entities or concepts (like computer file systems, family trees, etc). Graph nodes represent the concepts and interconnecting lines represent the relationships. We recorded participants' eye movements while viewing such graphs to test two possible models of graph comprehension. Graph readers had to answer interpretive questions, which required comparisons between two graph nodes. One model postulates a search and a combined search-reasoning stage of graph comprehension (two-stage model), whereas the second model predicts three stages, two stages devoted to the search of the relevant graph nodes and a separate reasoning stage. A detailed analysis of the eye movement data provided clear support for the three-stage model. This is in line with recent studies, which suggest that participants serialize problem solving tasks in order to minimize the overall processing load.

Eye Movements Indicate the Temporal Organisation of Information Processing in Graph Comprehension

“…These results indicated that reasoning processes that are essential for graph comprehension had not finished (or not even started) at the time when search was completed. These results are consistent with findings by Körner (2004) who observed that graph readers produced a substantial number of additional fixations after the second target had been located. Therefore, this behaviour suggests a strong serial component in graph comprehension.…”

“…For example, Huang et al (2009) used graphs with at most 22 nodes and 23 lines, and Körner (2004) and Körner and Albert (2002) had graphs with only nine nodes and 10 lines, whereas all the graphs of the present experiments included 75 nodes and 108 lines. It may seem plausible that there exists an interaction between the complexity of graphs and the amount of parallel processing that is possible during early stages of comprehension.…”

“…The comprehension of hierarchical graphs requires two fundamental processes: search and reasoning (Körner, 2004). Search is the localisation of the relevant nodes in the graph.…”

“…Using eye-movement recordings, Körner (2004) demonstrated that participants produced a substantial number of additional fixations after the second target had been located. This suggested that graph processing continued after the search for the target nodes was completed.…”

Decisiones Estratégicas De Lectura Y Rendimiento en Tareas De Competencia Lectora Similares a Pisa

“…No obstante, aunque los textos no-continuos propicien una interpretación de la información más global, también el procesamiento de información icónica podría requerir procesos secuenciales (Carpenter y Shah, 1998;Pinker, 1990;Shah, 2001;Körner, 2004) con una dificultad creciente (Pozo y Postigo, 2000). Cuando un lector comprende, por ejemplo, un gráfico de ejes cartesianos, en primer lugar codifica parte del patrón visual e identifica las características más relevantes del mismo (p.…”