An eye movement study of insight problem solving.

The representational change theory of insight claims that insight problems cause impasses because they mislead problem solvers into constructing inappropriate initial representations. Insight is attained when the initial representation is changed. In the present study (N = 24), we tested three specific implications of these hypotheses against eye movements recorded while participants solved matchstick arithmetic problems. The results were consistent with the predictions, providing converging evidence with prior findings using solution rates and solution times. Alternative theories of insight can explain individual findings, but only the representational change theory accounts for both the performance data and the eye movement data. The present study also suggests that eye movement recordings provide an important new window into processes of insight problem solving.

Eye movement control in reading: a comparison of two types of models.

Two classes of models have been proposed to account for eye movement control during reading. Proponents of the 1st class of model claim that the decision of when to move the eyes (reflected in fixation duration) is primarily influenced by the status of on-line language processing such as lexical access. Supporters of the 2nd class of model, however, maintain that (a) lower level oculomotor factors such as fixation location govern the decision of when to move the eyes and (b) lexical variables exert only a weak influence. In this study, fixation duration on low-and high-frequency target words was examined as a function of fixation location and the number of fixations on a target word. The data are inconsistent with an oculomotor model.

Eye movement control in reading and visual search: Effects of word frequency.

Eye movements were recorded as subjects either read text or searched through texts for a target word. In the reading task, there was a robust word frequency effect wherein readers looked longer at low-frequency words than at high-frequency words. However, there was no frequency effect in the search task. The results suggest that decisions to move the eyes during reading are made on a different basis than they are during visual search. Implications for current models of eye movement control in reading are discussed.

Word frequency effects and eye movements during two readings of a text.

The present study examined the influence of word frequency on rereading performance. Subjects read short passages, each twice in succession, while their eye movements were monitored. During first presentations, each passage contained a target word of low or high frequency; during second presentations, the targets were either repeated or replaced by synonyms. In general, during the second readings readers made shorter duration fixations, fewer fixations, and longer saccades. When fixation times on the target words were examined, results showed that fixation durations were shorter for high frequency words during both readings and that the decrease in fixation duration was similar in magnitude for low and high frequency words. This suggests that word frequency and repetition independently influenced reading time. In addition, replacing a target with a synonym did not increase processing time for the replacement word. This suggests that conceptual repetition was sufficient for obtaining repetition effects when reading text.

The use of information below fixation in reading and in visual search.

Two experiments examined whether or not readers obtain useful information from below the currently fixated line. In Experiment 1, subjects read passages of text and the availability of visual information below the line fixated was manipulated using a variant of the moving window technique. Reading was no slower when there was no letter information below the fixated line than when there was full information below the fixated line. However, a condition which made the strings of letters below the fixated line less "wordlike" caused reading to be slowed down by about 6% relative to the other conditions. In Experiment 2, subjects searched for a target word through passages of text. Subjects occasionally detected targets below the line they were fixating; however, there was no clear evidence that the availability of information below the line made search more efficient. It thus appears that in reading (or in other tasks where words are being identified and the eyes move horizontally along a line of text), little visual information is extracted below the line of text fixated.

Monitoring changes in cognitive load during reading: an event-related brain potential and reaction time analysis.

Factors that contribute to cognitive load during reading were examined using a secondary task procedure. In three experiments, subjects read sets of passages twice in succession while auditory probes were presented. The N1-P2 and P300 components of the event-related brain potential and reaction time (RT) responses to secondary auditory probes were used as measures of load. N1-P2 responses indicated decreased load during the second reading, whereas P300 and RT responses indicated increased load during the second reading. The results are interpreted as reflecting changes in task demands. Lower level elements of the task, such as word recognition and local aspects of comprehension, required fewer resources during the second reading. The N1-P2 reflected this reduction in resource demands. By contrast, the amount of resources devoted to higher level processes, such as comparing the text with one's prior representation and updating memory, increased during the second reading. This resulted from task demands, which emphasized memory of the material. P300 and RT reflected this increase in higher level demands. Results are described in terms of attentional and task demands and are taken as support for a componential description of reading and task difficulty.

Knowledge, concepts, and inferences in childhood.

The role of knowledge in children's inferences was investigated in 3 experiments. Experiment 1 examined developmental changes in the role of categorical membership, perceptual appearance, and item complexity in inferences for natural kind and artifact concepts. Preschoolers (5-year-olds), second graders (8-year-olds), and fourth graders (10-year-olds) were taught novel properties about target concepts and asked whether each of 4 probes had those properties. Probes varied in category membership and perceptual appearance relative to the target item. Item complexity also varied. Experiments 2 and 3 examined inferences with known and unknown concepts for familiar and unfamiliar properties. Older children's knowledge led to differential weighting of categorical information over appearance but only for known concepts and/or familiar properties. Preschoolers made no distinction between category and appearance for either known or unknown concepts. Additionally, as target item complexity increased, older children made more inferences than preschoolers. No differences between inferences about natural kind and artifact concepts were found. The role of theories and knowledge in children's drawing of inferences is discussed.