New neuroscience research upends traditional cognitive models of reading

How does your brain transform written words into spoken ones in mere milliseconds? A new study published in The Journal of Neuroscience has found that a key brain region traditionally associated with speech production is engaged in reading far earlier than expected. Using targeted brain stimulation, researchers demonstrated that the left posterior inferior frontal cortex (pIFC) is essential for translating written words into spoken language within just 100 milliseconds after seeing a word—well before traditional models suggest.

For decades, scientists have sought to understand how the brain reads, particularly the sequence of events that turn written text into spoken words. Traditional models propose a “serial cascade,” where written words are processed in stages: visual recognition in the fusiform gyrus, phonological conversion in the supramarginal gyrus, and speech production in the pIFC. This sequence implies that each stage waits for input from the previous one.

However, recent neuroimaging studies show simultaneous activation of these regions during reading, raising questions about whether they operate independently or interact directly. The researchers aimed to clarify the role of the pIFC in reading by using transcranial magnetic stimulation (TMS), a non-invasive technique that temporarily disrupts brain activity.

“Traditional cognitive models of reading assume that speech production occurs after initial visual and phonological processing of written words,” explained study author Kimihiro Nakamura, the principal investigator at the Systems Neuroscience Section at the National Rehabilitation Center Research Institute.

“This seems a plausible and reasonable a priori assumption, but a series of more recent magnetoencephalography (MEG) studies show that the pIFC, classically associated with spoken production, responds to print at 100-150 ms after word-onset, almost simultaneously with posterior brain regions for visual and phonological processing. Moreover, the functional significance of this fast neural response is also unclear, because the left pIFC is now known to mediate different aspects of linguistic/non-linguistic processing. We therefore wanted to fill this gap between cognitive models and empirical data from functional brain imaging.”

In the study, 50 adults participated in two experiments. In the first experiment, participants performed three tasks: reading words aloud, making semantic judgments (deciding if a word referred to an animal or plant), and distinguishing the text’s color (a perceptual control task). The second experiment introduced an object-naming task to compare processes involved in reading to those used for general spoken language production.

During these tasks, TMS pulses were applied to each of the three brain regions at various time intervals: 50, 100, 150, and 200 milliseconds after participants were shown a written word. This precise timing allowed researchers to investigate when each brain region became active and whether disrupting its function affected task performance. Participants’ reaction times and accuracy were measured to determine the impact of TMS on their ability to perform each task.

The stimuli for the reading tasks consisted of words written in a phonologically regular script, meaning each character corresponded consistently to a sound. This choice minimized variability in how participants converted text into speech sounds, enabling researchers to isolate the specific contributions of each brain region.

The researchers found that the pIFC, long thought to act later in the reading process, played an early and critical role. When TMS was applied to the pIFC at 100 milliseconds after participants saw a written word, their ability to read aloud was impaired. This disruption was specific to reading and did not affect participants’ performance on the semantic or color-judgment tasks. These results suggest that the pIFC is directly involved in the rapid transformation of written words into speech sounds.

The fusiform gyrus also showed early involvement. Disrupting its function at 100 milliseconds impaired both reading and semantic tasks, highlighting its role in visual word recognition. Unlike the pIFC, however, the fusiform gyrus did not show a task-specific effect; its disruption affected tasks requiring both phonological and semantic processing.

“Most of the current knowledge of spatiotemporal dynamics in reading is derived from functional neuroimaging data with high-temporal resolution, such as ERP and MEG, according to which posterior brain systems responsible for visual and phonological processing respond to print at 250-500 ms after stimulus-onset,” Nakamura told PsyPost. “While the main goal of the study was to dissect the causal role of early pIFC activation in reading, our TMS results revealed that those other systems for reading also act much faster than assumed by most neurocognitive models of reading derived from ERP/MEG data. Because TMS is a brain stimulation method for transiently disrupting local neural activity, we argue that the observed gap could be attributed to possibles difference in timing between actual neuronal firing and peak response latencies estimated from ERP/MEG waveforms.”

The supramarginal gyrus displayed delayed activation, with TMS disrupting performance only at 150 milliseconds or later. This finding aligns with its established role in phonological processing, which occurs after initial visual recognition of words.

Experiment 2 further clarified the specificity of the pIFC’s role in reading. Participants performed both oral reading and object-naming tasks, with TMS applied at the same time intervals. Disrupting the pIFC impaired reading but had no effect on naming objects, even though both tasks required spoken responses. This suggests that the pIFC’s early activation during reading is tied to its role in converting text into speech sounds, rather than general speech production.

These findings challenge the long-held “serial cascade” model of reading, which posits that visual and phonological processing must be completed before speech production begins. Instead, the results suggest that the pIFC and fusiform gyrus process information in parallel, with the pIFC playing a key role in a “sublexical” pathway that rapidly connects visual and speech motor systems.

“Our TMS data provide the first causal evidence showing that the early activation of the left pIFC specifically mediates rapid generation of speech motor codes during reading, which probably relies on the enhanced long-distance connectivity between occipitotemporal and frontal cortices that developed with the extensive experience in reading,” Nakamura explained. “Our results also show that this region starts to respond to print approximately 30 ms faster than thought previously, but not necessarily in an ordered cascade as assumed by cognitive models of visual word processing.”

“While such direct and rudimentary neurocognitive pathway for print-to-sound conversion is known to help decipher text in children and people with brain damage, little is known about its role and status in proficient adult readers, who primarily rely on more effective whole-word recognition systems. In sum, we therefore suggest that the brain may have more resources than cognitive models believe – the seemingly dormant, fast sublexical pathway for pronunciation is fully functioning in literate adults.”

These findings not only deepen our understanding of how the brain handles reading but also have potential applications in addressing reading-related challenges, such as dyslexia. By identifying the early and critical role of the pIFC, researchers have opened new avenues for exploring how these pathways develop in literacy and how they might be enhanced through targeted interventions.

“We believe that the precise temporal dynamics during reading is of critical importance for understanding the neurophysiology of dyslexia and related disorders,” Nakamura said. “In this context, by combining such temporal dynamics information and high temporal resolution methods (e.g., EEG and electrical cortical stimulation), we are particularly interested in developing novel neuromodulation methodology for effective remediation and training dedicated to these disorders.”

“While early pIFC activation in reading was first documented in 2004 and reported by several subsequent MEG studies, its theoretical significance has still remained elusive, particularly because MEG allows only correlational inferences about structures and functions in the brain,” Nakamura added. “To address the issue, it is essential to identify behavioral effects arising when this particular activity is suppressed during reading. Because TMS can transiently disrupt the function of a given cortical structure in normal humans, our results resolved this remaining question and provide more compelling evidence by showing the causal link between early pIFC activation and behavior.”

The study, “Dissecting the causal role of early inferior frontal activation in reading,” was authored by Tomoki Uno, Kouji Takano, and Kimihiro Nakamura.