Reading

Pictures are often used as decorative elements in textual learning materials. They may be interesting and attention capturing, yet not necessarily needed for the understanding of the core learning content and thus are thought to function as so-called pictorial seductive details. Whether and why the use of decorative pictures in learning materials results in beneficial or detrimental effects on learning outcomes is still a matter of debate. The current study addressed the potential effect of decorative pictures on working memory load. Subjects (N=32) performed text reading tasks followed by n-back working memory tasks. In both task series the presence of decorative pictures (present/absent) as well as the working memory load (low/high) was manipulated. EEG and eye-tracking data was recorded during task performance. It was hypothesized that if the presence of decorative pictures would increase working memory load, this would be indicated by a decreased parietal alpha frequency band power and an increased pupil dilation. Furthermore, the effect might be modulated by the general working memory load induced by the tasks. The EEG alpha frequency band power and the pupil dilation showed a highly consistent pattern of results. Only for the n-back task the presence of decorative pictures led to a decrease in alpha frequency band power and an increase in pupil dilation. This effect did not interact with the general working memory load of the task. Behavioral performance measures showed no effects of added decorative pictures. I will discuss these results and planned follow-up studies.

How do we process visual information during natural reading? In a computational framework (evaluated based on single-word behavioral and brain data), we could previously show that redundant visual information of a script is "explained away" to optimize the amount of visual information to be processed. As a result, this model proposes that visual word recognition relies on an orthographic prediction error (oPE), which in turn depends on our orthographical knowledge. Here we investigate to which degree oPE representations are also involved in natural reading. In addition, we explore if high-level semantic prediction (sentence context), interacts with the 'lower-level' oPE. We measured eye movements during sentence reading from N=82 German native speakers and found that first fixation durations showed a significant oPE by predictability interaction. This interaction resulted from the presence of an oPE effect for non-predictable words while there was no identifiable effect of oPE for predictable words. This pattern indicates that the proposed oPE representations are only relevant when sentence-level semantic information cannot predict the upcoming word. In contrast, when words can be predicted at the sentence level, these higher-level and more specific predictions override the lower-level, context-free, and thus less specific predictions during visual-orthographic processing. We conclude that informationally optimized predictive processing based on orthographic knowledge is operative also in natural reading contexts, but only in the absence of more specific, higher-level (e.g., context-dependent) predictions.

Rapid Automatized Naming (RAN) is defined as the ability to name stimuli as fast as possible and they are used internationally for the early diagnosis of risk children with reading deficits. However, the relationship between RAN and reading is still a topic of critical debate and studies use RAN very differently regarding the condition (serial vs. discrete) and the stimuli (digits, letters, colours, objects). In this research, we implement both RAN conditions with stimuli controlled for (psycho)linguistic variables to explore the lexical-semantic processes in RAN. We tested 66 university students aged between 19 and 38 with both the serial and discrete RAN condition and controlled the stimuli for semantic category, word frequency, word length and name agreement. In addition, participants were tested on reading, executive functions (inhibition and shifting) and working memory. Results showed no frequency effect but a word length effect in both RAN conditions. No relationship between working memory performance and RAN was discovered, but we found a strong relationship between RAN and processing speed in the shifting task. Performance in word reading was correlated differently depending on lexical status and RAN condition. We discuss our findings in light of common and different lexical-semantic processes in RAN and reading, using the logogen model.

The hypothesis of reading time regularity states that the degree of regularity in measures of the reading process is informative about reading fluency and comprehension. The current study aims at testing this assumption, namely that eye movement fluctuations contingent on linguistic information differ in their temporal structure from endogenous fluctuations of eye movements that are not contingent on external information. To that end, three language-unrelated conditions were chosen which serve as ‘baselines’ for eye movements in the absence of external information (looking at blank screens, fixation crosses or random patterns of circles on a screen). Another three conditions were selected reflecting different degrees of available linguistic information (encoding x-sequences, reading scrambled texts, actual reading of newspaper articles). Eye movements of 25 native speakers of German were recorded with a sampling rate of 1000 Hz. Gaze steps were computed by differencing the raw 2D position data, and subsequently subjected to recurrence quantification analysis (RQA) quantifying various dynamic properties of the time series related to the degree of randomness and structure of their temporal evolution. The results show that eye movement fluctuations during text reading differ systematically in the strength and degree of temporal structure compared to ‘baseline’ conditions that putatively capture endogenous fluctuations of eye movements in the absence of (linguistic) information. These findings provide a new and important perspective for further studies investigating natural reading as complex, dynamical process using measures of temporal structure.