The Role of Occipitotemporal Network in Rapid Reading - An fMRI-based Study

Background

Reading is the most effective cognitive means for humans to acquire new knowledge. Although most native English-speaking adults read at an average speed of about 200 to 400 words per minute (w/min), many people hope to increase their reading speed to acquire new knowledge faster and more efficiently. Some speed reading enthusiasts claim they can read English sentences at speeds of 30,000 to 40,000 words per minute. To improve reading speed, various strategies may be adopted, such as suppressing pronunciation habits, focusing on key words or concepts while reading, avoiding re-reading sentences, and expanding the range of vision through training.

The proposal of these strategies has sparked research on brain activity during the reading process. Existing studies have shown that the reading process involves multiple brain regions of complex language processing. In particular, regions located in the visual cortex and temporal lobe are thought to play a key role in language processing and the regulation of reading speed. However, rapid reading is not just a simple sequential process, but involves dynamic interactions between multiple brain regions. However, early studies on these interactions were limited, so this study explores the role of the visual temporal lobe network in rapid reading through fMRI and advanced data analysis methods.

Paper Information

This paper was written by Dexin Sun, Zhilin Zhang, Naoya Oishi, Qi Dai, Dinh Ha Duy Thuy, Nobuhito Abe, Jun Tachibana, Shintaro Funahashi, Jinglong Wu, Toshiya Murai, and Hidenao Fukuyama, from institutions including the Shenzhen Institute of Advanced Technology and Kyoto University. The paper was published in “Neurosci. Bull.”, with DOI https://doi.org/10.1007/s12264-024-01251-w. The paper was received on April 23, 2023, and accepted on March 15, 2024.

Research Design

Research Process

This study initially recruited 23 native Japanese speakers who had participated in speed reading training courses, but 4 were unable to complete all experiments, so 19 participants were ultimately included (7 females, 12 males; age range 20 to 46 years, average 32.4 years). Participants performed reading tasks at different speeds (slow, medium, and fast) and were scanned under a 3 Tesla fMRI scanner.

Each fMRI experiment contained a total of 19 blocks (9 task blocks and 10 rest blocks), with each task block being a 30-second video presented at slow, medium, and fast speeds. Participants read sentences during task blocks and fixated on a cross on the screen during rest blocks. After the experiment, participants needed to answer content-related questions to assess their speed reading performance.

Data Processing

All data were preprocessed and analyzed using SPM12. Each participant’s echo-planar imaging (EPI) images were first corrected for geometric distortion, then spatially normalized using the standard Montreal Neurological Institute (MNI) template, and the EPI images were smoothed with a Gaussian kernel. The experiment also included high-pass filtering and low-frequency noise removal of brain functional data.

Whole-brain Activation Analysis and Parametric Modulation Analysis

Experimental data were modeled and analyzed at individual and group levels to explore the role of reading speed in changes in brain activation. In parametric modulation analysis, the modulation effect of reading speed on brain segment activation was studied by adjusting reading speed (slow, medium, fast).

Psychophysiological Interaction Analysis and Dynamic Causal Modeling Analysis

The study further explored functional connectivity and effective connectivity between different brain regions in the visual temporal lobe network during speed reading through psychophysiological interaction (PPI) analysis and dynamic causal modeling (DCM) analysis. VOT (ventral occipitotemporal cortex) was selected as the seed region to analyze functional connectivity in the speed reading training group and the control group.

Main Results

Behavioral Results

In the speed reading training group, the accuracy rates for slow reading, medium speed reading, and fast reading were 69.1%, 60.9%, and 63.1% respectively, with no significant difference. Overall, the accuracy of rapidly presented sentences was comparable to that of slowly presented sentences.

Whole-brain Activation Results

As reading speed increased, activation weakened in areas such as the left STS (superior temporal sulcus), left middle temporal gyrus, left middle occipital gyrus, and left precentral gyrus. The study showed that increased reading speed brings greater cognitive load and requires higher levels of spatial processing, but these areas become more loosely active.

Parametric Modulation Analysis

In the speed-negative contrast, the training group showed significant activation in areas including the left middle occipital gyrus and superior temporal gyrus, while the comparison group showed activation in the superior temporal gyrus and superior occipital gyrus. In the speed-positive contrast, activation areas were relatively sparse in both the training group and the comparison group, mainly including the middle occipital gyrus. Overall, the VOT, STS, and IO in the speed reading training group showed more significant inhibitory effects with increasing reading speed compared to the comparison group.

Functional Connectivity Analysis

The study found differences in functional connectivity at different reading speeds when VOT was used as the seed region. In particular, the weakening of functional connectivity in the left occipitotemporal network was most significant under fast reading conditions. In the training group, the activation patterns of VOT with STS and IO showed stronger connectivity changes in slow versus fast conditions, demonstrating the significant impact of speed reading training on connectivity patterns.

Dynamic Causal Modeling Analysis

DCM analysis revealed the modulation of effective connectivity. In the speed reading training group, stronger inhibitory connections from ASTS to VOT were found, while enhancement connections from IO to VOT were more significant. At the same time, the analysis showed that in the speed reading training group, the inhibitory activation effect of signals from ASTS to VOT was greater than in the comparison group, while the enhancement effect from IO to VOT was moderate, demonstrating the connection modulation effect of the brain’s visual temporal lobe network in speed reading training.

Conclusion

This study reveals the distributional changes in the language reading network related to speed reading training, indicating that changes in brain activity are associated with speed reading training:

1. In the reading experiment, participants mainly activated brain regions related to visual and reading processing, with specific rate increases leading to deactivation of the left occipitotemporal network.
2. Increased reading speed led to weakened functional connectivity of VOT and related regions in the left occipitotemporal pathway.
3. Using IO as sensory input, DCM results revealed effective connectivity in the occipitotemporal pathway, emphasizing the significance of understanding functional and effective connectivity in the language reading process.

This study provides a more precise mapping of brain functional changes during speed reading training, contributing to a deeper understanding of the brain’s language processing.